Google Git
Sign in
apache / hbase / 71c27818947aefc303a2ba16d22255a15b5f8998 / . / src / main / docbkx / book.xml
blob: 43af6ec11ffde03472cc0784b1358a92e7757a73 [file] [log] [blame]
<?xml version="1.0" encoding="UTF-8"?>
<!--
/**
*
* Licensed to the Apache Software Foundation (ASF) under one
* or more contributor license agreements. See the NOTICE file
* distributed with this work for additional information
* regarding copyright ownership. The ASF licenses this file
* to you under the Apache License, Version 2.0 (the
* "License"); you may not use this file except in compliance
* with the License. You may obtain a copy of the License at
*
* http://www.apache.org/licenses/LICENSE-2.0
*
* Unless required by applicable law or agreed to in writing, software
* distributed under the License is distributed on an "AS IS" BASIS,
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
* See the License for the specific language governing permissions and
* limitations under the License.
*/
-->
<book version="5.0" xmlns="http://docbook.org/ns/docbook"
xmlns:xlink="http://www.w3.org/1999/xlink"
xmlns:xi="http://www.w3.org/2001/XInclude"
xmlns:svg="http://www.w3.org/2000/svg"
xmlns:m="http://www.w3.org/1998/Math/MathML"
xmlns:html="http://www.w3.org/1999/xhtml"
xmlns:db="http://docbook.org/ns/docbook" xml:id="book">
<info>
<title><link xlink:href="http://www.hbase.org">
The Apache HBase&#153; Reference Guide
</link></title>
<subtitle><link xlink:href="http://www.hbase.org">
<inlinemediaobject>
<imageobject>
<imagedata align="middle" valign="middle" fileref="hbase_logo.png" />
</imageobject>
</inlinemediaobject>
</link>
</subtitle>
<copyright><year>2014</year><holder>Apache Software Foundation.
All Rights Reserved. Apache Hadoop, Hadoop, MapReduce, HDFS, Zookeeper, HBase, and the HBase project logo are trademarks of the Apache Software Foundation.
</holder>
</copyright>
<abstract>
<para>This is the official reference guide of
<link xlink:href="http://www.hbase.org">Apache HBase&#153;</link>,
a distributed, versioned, big data store built on top of
<link xlink:href="http://hadoop.apache.org/">Apache Hadoop&#153;</link> and
<link xlink:href="http://zookeeper.apache.org/">Apache ZooKeeper&#153;</link>.
</para>
</abstract>
<revhistory>
<revision>
<revnumber>
<?eval ${project.version}?>
</revnumber>
<date>
<?eval ${buildDate}?>
</date>
</revision>
</revhistory>
</info>
<!--XInclude some chapters-->
<xi:include xmlns:xi="http://www.w3.org/2001/XInclude" href="preface.xml" />
<xi:include xmlns:xi="http://www.w3.org/2001/XInclude" href="getting_started.xml" />
<xi:include xmlns:xi="http://www.w3.org/2001/XInclude" href="configuration.xml" />
<xi:include xmlns:xi="http://www.w3.org/2001/XInclude" href="upgrading.xml" />
<xi:include xmlns:xi="http://www.w3.org/2001/XInclude" href="shell.xml" />
<chapter xml:id="datamodel">
<title>Data Model</title>
<para>In short, applications store data into an HBase table.
Tables are made of rows and columns.
All columns in HBase belong to a particular column family.
Table cells -- the intersection of row and column
coordinates -- are versioned.
A cell’s content is an uninterpreted array of bytes.
</para>
<para>Table row keys are also byte arrays so almost anything can
serve as a row key from strings to binary representations of longs or
even serialized data structures. Rows in HBase tables
are sorted by row key. The sort is byte-ordered. All table accesses are
via the table row key -- its primary key.
</para>
<section xml:id="conceptual.view"><title>Conceptual View</title>
<para>
The following example is a slightly modified form of the one on page
2 of the <link xlink:href="http://research.google.com/archive/bigtable.html">BigTable</link> paper.
There is a table called <varname>webtable</varname> that contains two column families named
<varname>contents</varname> and <varname>anchor</varname>.
In this example, <varname>anchor</varname> contains two
columns (<varname>anchor:cssnsi.com</varname>, <varname>anchor:my.look.ca</varname>)
and <varname>contents</varname> contains one column (<varname>contents:html</varname>).
<note>
<title>Column Names</title>
<para>
By convention, a column name is made of its column family prefix and a
<emphasis>qualifier</emphasis>. For example, the
column
<emphasis>contents:html</emphasis> is made up of the column family <varname>contents</varname>
and <varname>html</varname> qualifier.
The colon character (<literal
moreinfo="none">:</literal>) delimits the column family from the
column family <emphasis>qualifier</emphasis>.
</para>
</note>
<table frame='all'><title>Table <varname>webtable</varname></title>
<tgroup cols='4' align='left' colsep='1' rowsep='1'>
<colspec colname='c1'/>
<colspec colname='c2'/>
<colspec colname='c3'/>
<colspec colname='c4'/>
<thead>
<row><entry>Row Key</entry><entry>Time Stamp</entry><entry>ColumnFamily <varname>contents</varname></entry><entry>ColumnFamily <varname>anchor</varname></entry></row>
</thead>
<tbody>
<row><entry>"com.cnn.www"</entry><entry>t9</entry><entry></entry><entry><varname>anchor:cnnsi.com</varname> = "CNN"</entry></row>
<row><entry>"com.cnn.www"</entry><entry>t8</entry><entry></entry><entry><varname>anchor:my.look.ca</varname> = "CNN.com"</entry></row>
<row><entry>"com.cnn.www"</entry><entry>t6</entry><entry><varname>contents:html</varname> = "&lt;html&gt;..."</entry><entry></entry></row>
<row><entry>"com.cnn.www"</entry><entry>t5</entry><entry><varname>contents:html</varname> = "&lt;html&gt;..."</entry><entry></entry></row>
<row><entry>"com.cnn.www"</entry><entry>t3</entry><entry><varname>contents:html</varname> = "&lt;html&gt;..."</entry><entry></entry></row>
</tbody>
</tgroup>
</table>
</para>
</section>
<section xml:id="physical.view"><title>Physical View</title>
<para>
Although at a conceptual level tables may be viewed as a sparse set of rows.
Physically they are stored on a per-column family basis. New columns
(i.e., <varname>columnfamily:column</varname>) can be added to any
column family without pre-announcing them.
<table frame='all'><title>ColumnFamily <varname>anchor</varname></title>
<tgroup cols='3' align='left' colsep='1' rowsep='1'>
<colspec colname='c1'/>
<colspec colname='c2'/>
<colspec colname='c3'/>
<thead>
<row><entry>Row Key</entry><entry>Time Stamp</entry><entry>Column Family <varname>anchor</varname></entry></row>
</thead>
<tbody>
<row><entry>"com.cnn.www"</entry><entry>t9</entry><entry><varname>anchor:cnnsi.com</varname> = "CNN"</entry></row>
<row><entry>"com.cnn.www"</entry><entry>t8</entry><entry><varname>anchor:my.look.ca</varname> = "CNN.com"</entry></row>
</tbody>
</tgroup>
</table>
<table frame='all'><title>ColumnFamily <varname>contents</varname></title>
<tgroup cols='3' align='left' colsep='1' rowsep='1'>
<colspec colname='c1'/>
<colspec colname='c2'/>
<colspec colname='c3'/>
<thead>
<row><entry>Row Key</entry><entry>Time Stamp</entry><entry>ColumnFamily "contents:"</entry></row>
</thead>
<tbody>
<row><entry>"com.cnn.www"</entry><entry>t6</entry><entry><varname>contents:html</varname> = "&lt;html&gt;..."</entry></row>
<row><entry>"com.cnn.www"</entry><entry>t5</entry><entry><varname>contents:html</varname> = "&lt;html&gt;..."</entry></row>
<row><entry>"com.cnn.www"</entry><entry>t3</entry><entry><varname>contents:html</varname> = "&lt;html&gt;..."</entry></row>
</tbody>
</tgroup>
</table>
It is important to note in the diagram above that the empty cells shown in the
conceptual view are not stored since they need not be in a column-oriented
storage format. Thus a request for the value of the <varname>contents:html</varname>
column at time stamp <literal>t8</literal> would return no value. Similarly, a
request for an <varname>anchor:my.look.ca</varname> value at time stamp
<literal>t9</literal> would return no value. However, if no timestamp is
supplied, the most recent value for a particular column would be returned
and would also be the first one found since timestamps are stored in
descending order. Thus a request for the values of all columns in the row
<varname>com.cnn.www</varname> if no timestamp is specified would be:
the value of <varname>contents:html</varname> from time stamp
<literal>t6</literal>, the value of <varname>anchor:cnnsi.com</varname>
from time stamp <literal>t9</literal>, the value of
<varname>anchor:my.look.ca</varname> from time stamp <literal>t8</literal>.
</para>
<para>For more information about the internals of how Apache HBase stores data, see <xref linkend="regions.arch" />.
</para>
</section>
<section xml:id="namespace">
<title>Namespace</title>
<para>
A namespace is a logical grouping of tables analogous to a database in relation database
systems. This abstraction lays the groundwork for upcoming multi-tenancy related features:
<itemizedlist>
<listitem>Quota Management (HBASE-8410) - Restrict the amount of resources (ie
regions, tables) a namespace can consume.</listitem>
<listitem>Namespace Security Administration (HBASE-9206) - provide another
level of security administration for tenants.</listitem>
<listitem>Region server groups (HBASE-6721) - A namespace/table can be
pinned onto a subset of regionservers thus guaranteeing a course level of
isolation.</listitem>
</itemizedlist>
</para>
<section xml:id="namespace_creation">
<title>Namespace management</title>
<para>
A namespace can be created, removed or altered. Namespace membership is determined during
table creation by specifying a fully-qualified table name of the form:
<para>
<code>&lt;table namespace&gt;:&lt;table qualifier&gt;</code>
</para>
<para>
Examples:
</para>
<programlisting>
#Create a namespace
create_namespace 'my_ns'
#create my_table in my_ns namespace
create 'my_ns:my_table', 'fam'
#drop namespace
drop_namespace 'my_ns'
#alter namespace
alter_namespace 'my_ns', {METHOD => 'set', 'PROPERTY_NAME' => 'PROPERTY_VALUE'}
</programlisting>
</para>
</section>
<section xml:id="namespace_special">
<title>Predefined namespaces</title>
<para>
There are two predefined special namespaces:
<itemizedlist>
<listitem>hbase - system namespace, used to contain hbase internal tables</listitem>
<listitem>default - tables with no explicit specified namespace will automatically
fall into this namespace.</listitem>
</itemizedlist>
</para>
<para>
Examples:
<programlisting>
#namespace=foo and table qualifier=bar
create 'foo:bar', 'fam'
#namespace=default and table qualifier=bar
create 'bar', 'fam'
</programlisting>
</para>
</section>
</section>
<section xml:id="table">
<title>Table</title>
<para>
Tables are declared up front at schema definition time.
</para>
</section>
<section xml:id="row">
<title>Row</title>
<para>Row keys are uninterrpreted bytes. Rows are
lexicographically sorted with the lowest order appearing first
in a table. The empty byte array is used to denote both the
start and end of a tables' namespace.</para>
</section>
<section xml:id="columnfamily">
<title>Column Family<indexterm><primary>Column Family</primary></indexterm></title>
<para>
Columns in Apache HBase are grouped into <emphasis>column families</emphasis>.
All column members of a column family have the same prefix. For example, the
columns <emphasis>courses:history</emphasis> and
<emphasis>courses:math</emphasis> are both members of the
<emphasis>courses</emphasis> column family.
The colon character (<literal
moreinfo="none">:</literal>) delimits the column family from the
<indexterm>column family <emphasis>qualifier</emphasis><primary>Column Family Qualifier</primary></indexterm>.
The column family prefix must be composed of
<emphasis>printable</emphasis> characters. The qualifying tail, the
column family <emphasis>qualifier</emphasis>, can be made of any
arbitrary bytes. Column families must be declared up front
at schema definition time whereas columns do not need to be
defined at schema time but can be conjured on the fly while
the table is up an running.</para>
<para>Physically, all column family members are stored together on the
filesystem. Because tunings and
storage specifications are done at the column family level, it is
advised that all column family members have the same general access
pattern and size characteristics.</para>
<para></para>
</section>
<section xml:id="cells">
<title>Cells<indexterm><primary>Cells</primary></indexterm></title>
<para>A <emphasis>{row, column, version} </emphasis>tuple exactly
specifies a <literal>cell</literal> in HBase.
Cell content is uninterrpreted bytes</para>
</section>
<section xml:id="data_model_operations">
<title>Data Model Operations</title>
<para>The four primary data model operations are Get, Put, Scan, and Delete. Operations are applied via
<link xlink:href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/client/HTable.html">HTable</link> instances.
</para>
<section xml:id="get">
<title>Get</title>
<para><link xlink:href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/client/Get.html">Get</link> returns
attributes for a specified row. Gets are executed via
<link xlink:href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/client/HTable.html#get%28org.apache.hadoop.hbase.client.Get%29">
HTable.get</link>.
</para>
</section>
<section xml:id="put">
<title>Put</title>
<para><link xlink:href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/client/Put.html">Put</link> either
adds new rows to a table (if the key is new) or can update existing rows (if the key already exists). Puts are executed via
<link xlink:href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/client/HTable.html#put%28org.apache.hadoop.hbase.client.Put%29">
HTable.put</link> (writeBuffer) or <link xlink:href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/client/HTable.html#batch%28java.util.List%29">
HTable.batch</link> (non-writeBuffer).
</para>
</section>
<section xml:id="scan">
<title>Scans</title>
<para><link xlink:href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/client/Scan.html">Scan</link> allow
iteration over multiple rows for specified attributes.
</para>
<para>The following is an example of a
on an HTable table instance. Assume that a table is populated with rows with keys "row1", "row2", "row3",
and then another set of rows with the keys "abc1", "abc2", and "abc3". The following example shows how startRow and stopRow
can be applied to a Scan instance to return the rows beginning with "row".
<programlisting>
public static final byte[] CF = "cf".getBytes();
public static final byte[] ATTR = "attr".getBytes();
...
HTable htable = ... // instantiate HTable
Scan scan = new Scan();
scan.addColumn(CF, ATTR);
scan.setStartRow(Bytes.toBytes("row")); // start key is inclusive
scan.setStopRow(Bytes.toBytes("rox")); // stop key is exclusive
ResultScanner rs = htable.getScanner(scan);
try {
for (Result r = rs.next(); r != null; r = rs.next()) {
// process result...
} finally {
rs.close(); // always close the ResultScanner!
}
</programlisting>
</para>
<para>Note that generally the easiest way to specify a specific stop point for a scan is by using the <link xlink:href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/filter/InclusiveStopFilter.html">InclusiveStopFilter</link> class.
</para>
</section>
<section xml:id="delete">
<title>Delete</title>
<para><link xlink:href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/client/Delete.html">Delete</link> removes
a row from a table. Deletes are executed via
<link xlink:href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/client/HTable.html#delete%28org.apache.hadoop.hbase.client.Delete%29">
HTable.delete</link>.
</para>
<para>HBase does not modify data in place, and so deletes are handled by creating new markers called <emphasis>tombstones</emphasis>.
These tombstones, along with the dead values, are cleaned up on major compactions.
</para>
<para>See <xref linkend="version.delete"/> for more information on deleting versions of columns, and see
<xref linkend="compaction"/> for more information on compactions.
</para>
</section>
</section>
<section xml:id="versions">
<title>Versions<indexterm><primary>Versions</primary></indexterm></title>
<para>A <emphasis>{row, column, version} </emphasis>tuple exactly
specifies a <literal>cell</literal> in HBase. It's possible to have an
unbounded number of cells where the row and column are the same but the
cell address differs only in its version dimension.</para>
<para>While rows and column keys are expressed as bytes, the version is
specified using a long integer. Typically this long contains time
instances such as those returned by
<code>java.util.Date.getTime()</code> or
<code>System.currentTimeMillis()</code>, that is: <quote>the difference,
measured in milliseconds, between the current time and midnight, January
1, 1970 UTC</quote>.</para>
<para>The HBase version dimension is stored in decreasing order, so that
when reading from a store file, the most recent values are found
first.</para>
<para>There is a lot of confusion over the semantics of
<literal>cell</literal> versions, in HBase. In particular, a couple
questions that often come up are:<itemizedlist>
<listitem>
<para>If multiple writes to a cell have the same version, are all
versions maintained or just the last?<footnote>
<para>Currently, only the last written is fetchable.</para>
</footnote></para>
</listitem>
<listitem>
<para>Is it OK to write cells in a non-increasing version
order?<footnote>
<para>Yes</para>
</footnote></para>
</listitem>
</itemizedlist></para>
<para>Below we describe how the version dimension in HBase currently
works<footnote>
<para>See <link
xlink:href="https://issues.apache.org/jira/browse/HBASE-2406">HBASE-2406</link>
for discussion of HBase versions. <link
xlink:href="http://outerthought.org/blog/417-ot.html">Bending time
in HBase</link> makes for a good read on the version, or time,
dimension in HBase. It has more detail on versioning than is
provided here. As of this writing, the limiitation
<emphasis>Overwriting values at existing timestamps</emphasis>
mentioned in the article no longer holds in HBase. This section is
basically a synopsis of this article by Bruno Dumon.</para>
</footnote>.</para>
<section xml:id="versions.ops">
<title>Versions and HBase Operations</title>
<para>In this section we look at the behavior of the version dimension
for each of the core HBase operations.</para>
<section>
<title>Get/Scan</title>
<para>Gets are implemented on top of Scans. The below discussion of
<link xlink:href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/client/Get.html">Get</link> applies equally to <link
xlink:href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/client/Scan.html">Scans</link>.</para>
<para>By default, i.e. if you specify no explicit version, when
doing a <literal>get</literal>, the cell whose version has the
largest value is returned (which may or may not be the latest one
written, see later). The default behavior can be modified in the
following ways:</para>
<itemizedlist>
<listitem>
<para>to return more than one version, see <link
xlink:href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/client/Get.html#setMaxVersions()">Get.setMaxVersions()</link></para>
</listitem>
<listitem>
<para>to return versions other than the latest, see <link
xlink:href="???">Get.setTimeRange()</link></para>
<para>To retrieve the latest version that is less than or equal
to a given value, thus giving the 'latest' state of the record
at a certain point in time, just use a range from 0 to the
desired version and set the max versions to 1.</para>
</listitem>
</itemizedlist>
</section>
<section xml:id="default_get_example">
<title>Default Get Example</title>
<para>The following Get will only retrieve the current version of the row
<programlisting>
public static final byte[] CF = "cf".getBytes();
public static final byte[] ATTR = "attr".getBytes();
...
Get get = new Get(Bytes.toBytes("row1"));
Result r = htable.get(get);
byte[] b = r.getValue(CF, ATTR); // returns current version of value
</programlisting>
</para>
</section>
<section xml:id="versioned_get_example">
<title>Versioned Get Example</title>
<para>The following Get will return the last 3 versions of the row.
<programlisting>
public static final byte[] CF = "cf".getBytes();
public static final byte[] ATTR = "attr".getBytes();
...
Get get = new Get(Bytes.toBytes("row1"));
get.setMaxVersions(3); // will return last 3 versions of row
Result r = htable.get(get);
byte[] b = r.getValue(CF, ATTR); // returns current version of value
List&lt;KeyValue&gt; kv = r.getColumn(CF, ATTR); // returns all versions of this column
</programlisting>
</para>
</section>
<section>
<title>Put</title>
<para>Doing a put always creates a new version of a
<literal>cell</literal>, at a certain timestamp. By default the
system uses the server's <literal>currentTimeMillis</literal>, but
you can specify the version (= the long integer) yourself, on a
per-column level. This means you could assign a time in the past or
the future, or use the long value for non-time purposes.</para>
<para>To overwrite an existing value, do a put at exactly the same
row, column, and version as that of the cell you would
overshadow.</para>
<section xml:id="implicit_version_example">
<title>Implicit Version Example</title>
<para>The following Put will be implicitly versioned by HBase with the current time.
<programlisting>
public static final byte[] CF = "cf".getBytes();
public static final byte[] ATTR = "attr".getBytes();
...
Put put = new Put(Bytes.toBytes(row));
put.add(CF, ATTR, Bytes.toBytes( data));
htable.put(put);
</programlisting>
</para>
</section>
<section xml:id="explicit_version_example">
<title>Explicit Version Example</title>
<para>The following Put has the version timestamp explicitly set.
<programlisting>
public static final byte[] CF = "cf".getBytes();
public static final byte[] ATTR = "attr".getBytes();
...
Put put = new Put( Bytes.toBytes(row));
long explicitTimeInMs = 555; // just an example
put.add(CF, ATTR, explicitTimeInMs, Bytes.toBytes(data));
htable.put(put);
</programlisting>
Caution: the version timestamp is internally by HBase for things like time-to-live calculations.
It's usually best to avoid setting this timestamp yourself. Prefer using a separate
timestamp attribute of the row, or have the timestamp a part of the rowkey, or both.
</para>
</section>
</section>
<section xml:id="version.delete">
<title>Delete</title>
<para>There are three different types of internal delete markers
<footnote><para>See Lars Hofhansl's blog for discussion of his attempt
adding another, <link xlink:href="http://hadoop-hbase.blogspot.com/2012/01/scanning-in-hbase.html">Scanning in HBase: Prefix Delete Marker</link></para></footnote>:
<itemizedlist>
<listitem><para>Delete: for a specific version of a column.</para>
</listitem>
<listitem><para>Delete column: for all versions of a column.</para>
</listitem>
<listitem><para>Delete family: for all columns of a particular ColumnFamily</para>
</listitem>
</itemizedlist>
When deleting an entire row, HBase will internally create a tombstone for each ColumnFamily (i.e., not each individual column).
</para>
<para>Deletes work by creating <emphasis>tombstone</emphasis>
markers. For example, let's suppose we want to delete a row. For
this you can specify a version, or else by default the
<literal>currentTimeMillis</literal> is used. What this means is
<quote>delete all cells where the version is less than or equal to
this version</quote>. HBase never modifies data in place, so for
example a delete will not immediately delete (or mark as deleted)
the entries in the storage file that correspond to the delete
condition. Rather, a so-called <emphasis>tombstone</emphasis> is
written, which will mask the deleted values<footnote>
<para>When HBase does a major compaction, the tombstones are
processed to actually remove the dead values, together with the
tombstones themselves.</para>
</footnote>. If the version you specified when deleting a row is
larger than the version of any value in the row, then you can
consider the complete row to be deleted.</para>
<para>For an informative discussion on how deletes and versioning interact, see
the thread <link xlink:href="http://comments.gmane.org/gmane.comp.java.hadoop.hbase.user/28421">Put w/ timestamp -> Deleteall -> Put w/ timestamp fails</link>
up on the user mailing list.</para>
<para>Also see <xref linkend="keyvalue"/> for more information on the internal KeyValue format.
</para>
</section>
</section>
<section>
<title>Current Limitations</title>
<section>
<title>Deletes mask Puts</title>
<para>Deletes mask puts, even puts that happened after the delete
was entered<footnote>
<para><link
xlink:href="https://issues.apache.org/jira/browse/HBASE-2256">HBASE-2256</link></para>
</footnote>. Remember that a delete writes a tombstone, which only
disappears after then next major compaction has run. Suppose you do
a delete of everything &lt;= T. After this you do a new put with a
timestamp &lt;= T. This put, even if it happened after the delete,
will be masked by the delete tombstone. Performing the put will not
fail, but when you do a get you will notice the put did have no
effect. It will start working again after the major compaction has
run. These issues should not be a problem if you use
always-increasing versions for new puts to a row. But they can occur
even if you do not care about time: just do delete and put
immediately after each other, and there is some chance they happen
within the same millisecond.</para>
</section>
<section>
<title>Major compactions change query results</title>
<para><quote>...create three cell versions at t1, t2 and t3, with a
maximum-versions setting of 2. So when getting all versions, only
the values at t2 and t3 will be returned. But if you delete the
version at t2 or t3, the one at t1 will appear again. Obviously,
once a major compaction has run, such behavior will not be the case
anymore...<footnote>
<para>See <emphasis>Garbage Collection</emphasis> in <link
xlink:href="http://outerthought.org/blog/417-ot.html">Bending
time in HBase</link> </para>
</footnote></quote></para>
</section>
</section>
</section>
<section xml:id="dm.sort">
<title>Sort Order</title>
<para>All data model operations HBase return data in sorted order. First by row,
then by ColumnFamily, followed by column qualifier, and finally timestamp (sorted
in reverse, so newest records are returned first).
</para>
</section>
<section xml:id="dm.column.metadata">
<title>Column Metadata</title>
<para>There is no store of column metadata outside of the internal KeyValue instances for a ColumnFamily.
Thus, while HBase can support not only a wide number of columns per row, but a heterogenous set of columns
between rows as well, it is your responsibility to keep track of the column names.
</para>
<para>The only way to get a complete set of columns that exist for a ColumnFamily is to process all the rows.
For more information about how HBase stores data internally, see <xref linkend="keyvalue" />.
</para>
</section>
<section xml:id="joins"><title>Joins</title>
<para>Whether HBase supports joins is a common question on the dist-list, and there is a simple answer: it doesn't,
at not least in the way that RDBMS' support them (e.g., with equi-joins or outer-joins in SQL). As has been illustrated
in this chapter, the read data model operations in HBase are Get and Scan.
</para>
<para>However, that doesn't mean that equivalent join functionality can't be supported in your application, but
you have to do it yourself. The two primary strategies are either denormalizing the data upon writing to HBase,
or to have lookup tables and do the join between HBase tables in your application or MapReduce code (and as RDBMS'
demonstrate, there are several strategies for this depending on the size of the tables, e.g., nested loops vs.
hash-joins). So which is the best approach? It depends on what you are trying to do, and as such there isn't a single
answer that works for every use case.
</para>
</section>
<section xml:id="acid"><title>ACID</title>
<pre>See <link xlink:href="http://hbase.apache.org/acid-semantics.html">ACID Semantics</link>.
Lars Hofhansl has also written a note on
<link xlink:href="http://hadoop-hbase.blogspot.com/2012/03/acid-in-hbase.html">ACID in HBase</link>.</pre>
</section>
</chapter> <!-- data model -->
<!-- schema design -->
<xi:include xmlns:xi="http://www.w3.org/2001/XInclude" href="schema_design.xml" />
<chapter xml:id="mapreduce">
<title>HBase and MapReduce</title>
<para>See <link xlink:href="http://hbase.org/apidocs/org/apache/hadoop/hbase/mapreduce/package-summary.html#package_description">
HBase and MapReduce</link> up in javadocs.
Start there. Below is some additional help.</para>
<para>For more information about MapReduce (i.e., the framework in general), see the Hadoop site (TODO: Need good links here --
we used to have some but they rotted against apache hadoop).</para>
<caution>
<title>Notice to Mapreduce users of HBase 0.96.1 and above</title>
<para>Some mapreduce jobs that use HBase fail to launch. The symptom is an
exception similar to the following:
<programlisting>
Exception in thread "main" java.lang.IllegalAccessError: class
com.google.protobuf.ZeroCopyLiteralByteString cannot access its superclass
com.google.protobuf.LiteralByteString
at java.lang.ClassLoader.defineClass1(Native Method)
at java.lang.ClassLoader.defineClass(ClassLoader.java:792)
at java.security.SecureClassLoader.defineClass(SecureClassLoader.java:142)
at java.net.URLClassLoader.defineClass(URLClassLoader.java:449)
at java.net.URLClassLoader.access$100(URLClassLoader.java:71)
at java.net.URLClassLoader$1.run(URLClassLoader.java:361)
at java.net.URLClassLoader$1.run(URLClassLoader.java:355)
at java.security.AccessController.doPrivileged(Native Method)
at java.net.URLClassLoader.findClass(URLClassLoader.java:354)
at java.lang.ClassLoader.loadClass(ClassLoader.java:424)
at java.lang.ClassLoader.loadClass(ClassLoader.java:357)
at
org.apache.hadoop.hbase.protobuf.ProtobufUtil.toScan(ProtobufUtil.java:818)
at
org.apache.hadoop.hbase.mapreduce.TableMapReduceUtil.convertScanToString(TableMapReduceUtil.java:433)
at
org.apache.hadoop.hbase.mapreduce.TableMapReduceUtil.initTableMapperJob(TableMapReduceUtil.java:186)
at
org.apache.hadoop.hbase.mapreduce.TableMapReduceUtil.initTableMapperJob(TableMapReduceUtil.java:147)
at
org.apache.hadoop.hbase.mapreduce.TableMapReduceUtil.initTableMapperJob(TableMapReduceUtil.java:270)
at
org.apache.hadoop.hbase.mapreduce.TableMapReduceUtil.initTableMapperJob(TableMapReduceUtil.java:100)
...
</programlisting>
This is because of an optimization introduced in <link
xlink:href="https://issues.apache.org/jira/browse/HBASE-9867">HBASE-9867</link>
that inadvertently introduced a classloader dependency.
</para>
<para>This affects both jobs using the <code>-libjars</code> option and
"fat jar," those which package their runtime dependencies in a nested
<code>lib</code> folder.</para>
<para>In order to satisfy the new classloader requirements,
hbase-protocol.jar must be included in Hadoop's classpath. This can be
resolved system-wide by including a reference to the hbase-protocol.jar in
hadoop's lib directory, via a symlink or by copying the jar into the new
location.</para>
<para>This can also be achieved on a per-job launch basis by including it
in the <code>HADOOP_CLASSPATH</code> environment variable at job submission
time. When launching jobs that package their dependencies, all three of the
following job launching commands satisfy this requirement:</para>
<programlisting>
$ HADOOP_CLASSPATH=/path/to/hbase-protocol.jar:/path/to/hbase/conf hadoop jar MyJob.jar MyJobMainClass
$ HADOOP_CLASSPATH=$(hbase mapredcp):/path/to/hbase/conf hadoop jar MyJob.jar MyJobMainClass
$ HADOOP_CLASSPATH=$(hbase classpath) hadoop jar MyJob.jar MyJobMainClass
</programlisting>
<para>For jars that do not package their dependencies, the following command
structure is necessary:</para>
<programlisting>
$ HADOOP_CLASSPATH=$(hbase mapredcp):/etc/hbase/conf hadoop jar MyApp.jar MyJobMainClass -libjars $(hbase mapredcp | tr ':' ',') ...
</programlisting>
<para>See also <link
xlink:href="https://issues.apache.org/jira/browse/HBASE-10304">HBASE-10304</link>
for further discussion of this issue.</para>
</caution>
<section xml:id="splitter">
<title>Map-Task Splitting</title>
<section xml:id="splitter.default">
<title>The Default HBase MapReduce Splitter</title>
<para>When <link xlink:href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/mapreduce/TableInputFormat.html">TableInputFormat</link>
is used to source an HBase table in a MapReduce job,
its splitter will make a map task for each region of the table.
Thus, if there are 100 regions in the table, there will be
100 map-tasks for the job - regardless of how many column families are selected in the Scan.</para>
</section>
<section xml:id="splitter.custom">
<title>Custom Splitters</title>
<para>For those interested in implementing custom splitters, see the method <code>getSplits</code> in
<link xlink:href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/mapreduce/TableInputFormatBase.html">TableInputFormatBase</link>.
That is where the logic for map-task assignment resides.
</para>
</section>
</section>
<section xml:id="mapreduce.example">
<title>HBase MapReduce Examples</title>
<section xml:id="mapreduce.example.read">
<title>HBase MapReduce Read Example</title>
<para>The following is an example of using HBase as a MapReduce source in read-only manner. Specifically,
there is a Mapper instance but no Reducer, and nothing is being emitted from the Mapper. There job would be defined
as follows...
<programlisting>
Configuration config = HBaseConfiguration.create();
Job job = new Job(config, "ExampleRead");
job.setJarByClass(MyReadJob.class); // class that contains mapper
Scan scan = new Scan();
scan.setCaching(500); // 1 is the default in Scan, which will be bad for MapReduce jobs
scan.setCacheBlocks(false); // don't set to true for MR jobs
// set other scan attrs
...
TableMapReduceUtil.initTableMapperJob(
tableName, // input HBase table name
scan, // Scan instance to control CF and attribute selection
MyMapper.class, // mapper
null, // mapper output key
null, // mapper output value
job);
job.setOutputFormatClass(NullOutputFormat.class); // because we aren't emitting anything from mapper
boolean b = job.waitForCompletion(true);
if (!b) {
throw new IOException("error with job!");
}
</programlisting>
...and the mapper instance would extend <link xlink:href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/mapreduce/TableMapper.html">TableMapper</link>...
<programlisting>
public static class MyMapper extends TableMapper&lt;Text, Text&gt; {
public void map(ImmutableBytesWritable row, Result value, Context context) throws InterruptedException, IOException {
// process data for the row from the Result instance.
}
}
</programlisting>
</para>
</section>
<section xml:id="mapreduce.example.readwrite">
<title>HBase MapReduce Read/Write Example</title>
<para>The following is an example of using HBase both as a source and as a sink with MapReduce.
This example will simply copy data from one table to another.
<programlisting>
Configuration config = HBaseConfiguration.create();
Job job = new Job(config,"ExampleReadWrite");
job.setJarByClass(MyReadWriteJob.class); // class that contains mapper
Scan scan = new Scan();
scan.setCaching(500); // 1 is the default in Scan, which will be bad for MapReduce jobs
scan.setCacheBlocks(false); // don't set to true for MR jobs
// set other scan attrs
TableMapReduceUtil.initTableMapperJob(
sourceTable, // input table
scan, // Scan instance to control CF and attribute selection
MyMapper.class, // mapper class
null, // mapper output key
null, // mapper output value
job);
TableMapReduceUtil.initTableReducerJob(
targetTable, // output table
null, // reducer class
job);
job.setNumReduceTasks(0);
boolean b = job.waitForCompletion(true);
if (!b) {
throw new IOException("error with job!");
}
</programlisting>
An explanation is required of what <classname>TableMapReduceUtil</classname> is doing, especially with the reducer.
<link xlink:href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/mapreduce/TableOutputFormat.html">TableOutputFormat</link> is being used
as the outputFormat class, and several parameters are being set on the config (e.g., TableOutputFormat.OUTPUT_TABLE), as
well as setting the reducer output key to <classname>ImmutableBytesWritable</classname> and reducer value to <classname>Writable</classname>.
These could be set by the programmer on the job and conf, but <classname>TableMapReduceUtil</classname> tries to make things easier.
<para>The following is the example mapper, which will create a <classname>Put</classname> and matching the input <classname>Result</classname>
and emit it. Note: this is what the CopyTable utility does.
</para>
<programlisting>
public static class MyMapper extends TableMapper&lt;ImmutableBytesWritable, Put&gt; {
public void map(ImmutableBytesWritable row, Result value, Context context) throws IOException, InterruptedException {
// this example is just copying the data from the source table...
context.write(row, resultToPut(row,value));
}
private static Put resultToPut(ImmutableBytesWritable key, Result result) throws IOException {
Put put = new Put(key.get());
for (KeyValue kv : result.raw()) {
put.add(kv);
}
return put;
}
}
</programlisting>
<para>There isn't actually a reducer step, so <classname>TableOutputFormat</classname> takes care of sending the <classname>Put</classname>
to the target table.
</para>
<para>This is just an example, developers could choose not to use <classname>TableOutputFormat</classname> and connect to the
target table themselves.
</para>
</para>
</section>
<section xml:id="mapreduce.example.readwrite.multi">
<title>HBase MapReduce Read/Write Example With Multi-Table Output</title>
<para>TODO: example for <classname>MultiTableOutputFormat</classname>.
</para>
</section>
<section xml:id="mapreduce.example.summary">
<title>HBase MapReduce Summary to HBase Example</title>
<para>The following example uses HBase as a MapReduce source and sink with a summarization step. This example will
count the number of distinct instances of a value in a table and write those summarized counts in another table.
<programlisting>
Configuration config = HBaseConfiguration.create();
Job job = new Job(config,"ExampleSummary");
job.setJarByClass(MySummaryJob.class); // class that contains mapper and reducer
Scan scan = new Scan();
scan.setCaching(500); // 1 is the default in Scan, which will be bad for MapReduce jobs
scan.setCacheBlocks(false); // don't set to true for MR jobs
// set other scan attrs
TableMapReduceUtil.initTableMapperJob(
sourceTable, // input table
scan, // Scan instance to control CF and attribute selection
MyMapper.class, // mapper class
Text.class, // mapper output key
IntWritable.class, // mapper output value
job);
TableMapReduceUtil.initTableReducerJob(
targetTable, // output table
MyTableReducer.class, // reducer class
job);
job.setNumReduceTasks(1); // at least one, adjust as required
boolean b = job.waitForCompletion(true);
if (!b) {
throw new IOException("error with job!");
}
</programlisting>
In this example mapper a column with a String-value is chosen as the value to summarize upon.
This value is used as the key to emit from the mapper, and an <classname>IntWritable</classname> represents an instance counter.
<programlisting>
public static class MyMapper extends TableMapper&lt;Text, IntWritable&gt; {
public static final byte[] CF = "cf".getBytes();
public static final byte[] ATTR1 = "attr1".getBytes();
private final IntWritable ONE = new IntWritable(1);
private Text text = new Text();
public void map(ImmutableBytesWritable row, Result value, Context context) throws IOException, InterruptedException {
String val = new String(value.getValue(CF, ATTR1));
text.set(val); // we can only emit Writables...
context.write(text, ONE);
}
}
</programlisting>
In the reducer, the "ones" are counted (just like any other MR example that does this), and then emits a <classname>Put</classname>.
<programlisting>
public static class MyTableReducer extends TableReducer&lt;Text, IntWritable, ImmutableBytesWritable&gt; {
public static final byte[] CF = "cf".getBytes();
public static final byte[] COUNT = "count".getBytes();
public void reduce(Text key, Iterable&lt;IntWritable&gt; values, Context context) throws IOException, InterruptedException {
int i = 0;
for (IntWritable val : values) {
i += val.get();
}
Put put = new Put(Bytes.toBytes(key.toString()));
put.add(CF, COUNT, Bytes.toBytes(i));
context.write(null, put);
}
}
</programlisting>
</para>
</section>
<section xml:id="mapreduce.example.summary.file">
<title>HBase MapReduce Summary to File Example</title>
<para>This very similar to the summary example above, with exception that this is using HBase as a MapReduce source
but HDFS as the sink. The differences are in the job setup and in the reducer. The mapper remains the same.
</para>
<programlisting>
Configuration config = HBaseConfiguration.create();
Job job = new Job(config,"ExampleSummaryToFile");
job.setJarByClass(MySummaryFileJob.class); // class that contains mapper and reducer
Scan scan = new Scan();
scan.setCaching(500); // 1 is the default in Scan, which will be bad for MapReduce jobs
scan.setCacheBlocks(false); // don't set to true for MR jobs
// set other scan attrs
TableMapReduceUtil.initTableMapperJob(
sourceTable, // input table
scan, // Scan instance to control CF and attribute selection
MyMapper.class, // mapper class
Text.class, // mapper output key
IntWritable.class, // mapper output value
job);
job.setReducerClass(MyReducer.class); // reducer class
job.setNumReduceTasks(1); // at least one, adjust as required
FileOutputFormat.setOutputPath(job, new Path("/tmp/mr/mySummaryFile")); // adjust directories as required
boolean b = job.waitForCompletion(true);
if (!b) {
throw new IOException("error with job!");
}
</programlisting>
As stated above, the previous Mapper can run unchanged with this example.
As for the Reducer, it is a "generic" Reducer instead of extending TableMapper and emitting Puts.
<programlisting>
public static class MyReducer extends Reducer&lt;Text, IntWritable, Text, IntWritable&gt; {
public void reduce(Text key, Iterable&lt;IntWritable&gt; values, Context context) throws IOException, InterruptedException {
int i = 0;
for (IntWritable val : values) {
i += val.get();
}
context.write(key, new IntWritable(i));
}
}
</programlisting>
</section>
<section xml:id="mapreduce.example.summary.noreducer">
<title>HBase MapReduce Summary to HBase Without Reducer</title>
<para>It is also possible to perform summaries without a reducer - if you use HBase as the reducer.
</para>
<para>An HBase target table would need to exist for the job summary. The HTable method <code>incrementColumnValue</code>
would be used to atomically increment values. From a performance perspective, it might make sense to keep a Map
of values with their values to be incremeneted for each map-task, and make one update per key at during the <code>
cleanup</code> method of the mapper. However, your milage may vary depending on the number of rows to be processed and
unique keys.
</para>
<para>In the end, the summary results are in HBase.
</para>
</section>
<section xml:id="mapreduce.example.summary.rdbms">
<title>HBase MapReduce Summary to RDBMS</title>
<para>Sometimes it is more appropriate to generate summaries to an RDBMS. For these cases, it is possible
to generate summaries directly to an RDBMS via a custom reducer. The <code>setup</code> method
can connect to an RDBMS (the connection information can be passed via custom parameters in the context) and the
cleanup method can close the connection.
</para>
<para>It is critical to understand that number of reducers for the job affects the summarization implementation, and
you'll have to design this into your reducer. Specifically, whether it is designed to run as a singleton (one reducer)
or multiple reducers. Neither is right or wrong, it depends on your use-case. Recognize that the more reducers that
are assigned to the job, the more simultaneous connections to the RDBMS will be created - this will scale, but only to a point.
</para>
<programlisting>
public static class MyRdbmsReducer extends Reducer&lt;Text, IntWritable, Text, IntWritable&gt; {
private Connection c = null;
public void setup(Context context) {
// create DB connection...
}
public void reduce(Text key, Iterable&lt;IntWritable&gt; values, Context context) throws IOException, InterruptedException {
// do summarization
// in this example the keys are Text, but this is just an example
}
public void cleanup(Context context) {
// close db connection
}
}
</programlisting>
<para>In the end, the summary results are written to your RDBMS table/s.
</para>
</section>
</section> <!-- mr examples -->
<section xml:id="mapreduce.htable.access">
<title>Accessing Other HBase Tables in a MapReduce Job</title>
<para>Although the framework currently allows one HBase table as input to a
MapReduce job, other HBase tables can
be accessed as lookup tables, etc., in a
MapReduce job via creating an HTable instance in the setup method of the Mapper.
<programlisting>public class MyMapper extends TableMapper&lt;Text, LongWritable&gt; {
private HTable myOtherTable;
public void setup(Context context) {
myOtherTable = new HTable("myOtherTable");
}
public void map(ImmutableBytesWritable row, Result value, Context context) throws IOException, InterruptedException {
// process Result...
// use 'myOtherTable' for lookups
}
</programlisting>
</para>
</section>
<section xml:id="mapreduce.specex">
<title>Speculative Execution</title>
<para>It is generally advisable to turn off speculative execution for
MapReduce jobs that use HBase as a source. This can either be done on a
per-Job basis through properties, on on the entire cluster. Especially
for longer running jobs, speculative execution will create duplicate
map-tasks which will double-write your data to HBase; this is probably
not what you want.
</para>
<para>See <xref linkend="spec.ex"/> for more information.
</para>
</section>
</chapter> <!-- mapreduce -->
<xi:include xmlns:xi="http://www.w3.org/2001/XInclude" href="security.xml" />
<chapter xml:id="architecture">
<title>Architecture</title>
<section xml:id="arch.overview">
<title>Overview</title>
<section xml:id="arch.overview.nosql">
<title>NoSQL?</title>
<para>HBase is a type of "NoSQL" database. "NoSQL" is a general term meaning that the database isn't an RDBMS which
supports SQL as its primary access language, but there are many types of NoSQL databases: BerkeleyDB is an
example of a local NoSQL database, whereas HBase is very much a distributed database. Technically speaking,
HBase is really more a "Data Store" than "Data Base" because it lacks many of the features you find in an RDBMS,
such as typed columns, secondary indexes, triggers, and advanced query languages, etc.
</para>
<para>However, HBase has many features which supports both linear and modular scaling. HBase clusters expand
by adding RegionServers that are hosted on commodity class servers. If a cluster expands from 10 to 20
RegionServers, for example, it doubles both in terms of storage and as well as processing capacity.
RDBMS can scale well, but only up to a point - specifically, the size of a single database server - and for the best
performance requires specialized hardware and storage devices. HBase features of note are:
<itemizedlist>
<listitem>Strongly consistent reads/writes: HBase is not an "eventually consistent" DataStore. This
makes it very suitable for tasks such as high-speed counter aggregation. </listitem>
<listitem>Automatic sharding: HBase tables are distributed on the cluster via regions, and regions are
automatically split and re-distributed as your data grows.</listitem>
<listitem>Automatic RegionServer failover</listitem>
<listitem>Hadoop/HDFS Integration: HBase supports HDFS out of the box as its distributed file system.</listitem>
<listitem>MapReduce: HBase supports massively parallelized processing via MapReduce for using HBase as both
source and sink.</listitem>
<listitem>Java Client API: HBase supports an easy to use Java API for programmatic access.</listitem>
<listitem>Thrift/REST API: HBase also supports Thrift and REST for non-Java front-ends.</listitem>
<listitem>Block Cache and Bloom Filters: HBase supports a Block Cache and Bloom Filters for high volume query optimization.</listitem>
<listitem>Operational Management: HBase provides build-in web-pages for operational insight as well as JMX metrics.</listitem>
</itemizedlist>
</para>
</section>
<section xml:id="arch.overview.when">
<title>When Should I Use HBase?</title>
<para>HBase isn't suitable for every problem.</para>
<para>First, make sure you have enough data. If you have hundreds of millions or billions of rows, then
HBase is a good candidate. If you only have a few thousand/million rows, then using a traditional RDBMS
might be a better choice due to the fact that all of your data might wind up on a single node (or two) and
the rest of the cluster may be sitting idle.
</para>
<para>Second, make sure you can live without all the extra features that an RDBMS provides (e.g., typed columns,
secondary indexes, transactions, advanced query languages, etc.) An application built against an RDBMS cannot be
"ported" to HBase by simply changing a JDBC driver, for example. Consider moving from an RDBMS to HBase as a
complete redesign as opposed to a port.
</para>
<para>Third, make sure you have enough hardware. Even HDFS doesn't do well with anything less than
5 DataNodes (due to things such as HDFS block replication which has a default of 3), plus a NameNode.
</para>
<para>HBase can run quite well stand-alone on a laptop - but this should be considered a development
configuration only.
</para>
</section>
<section xml:id="arch.overview.hbasehdfs">
<title>What Is The Difference Between HBase and Hadoop/HDFS?</title>
<para><link xlink:href="http://hadoop.apache.org/hdfs/">HDFS</link> is a distributed file system that is well suited for the storage of large files.
Its documentation states that it is not, however, a general purpose file system, and does not provide fast individual record lookups in files.
HBase, on the other hand, is built on top of HDFS and provides fast record lookups (and updates) for large tables.
This can sometimes be a point of conceptual confusion. HBase internally puts your data in indexed "StoreFiles" that exist
on HDFS for high-speed lookups. See the <xref linkend="datamodel" /> and the rest of this chapter for more information on how HBase achieves its goals.
</para>
</section>
</section>
<section xml:id="arch.catalog">
<title>Catalog Tables</title>
<para>The catalog tables -ROOT- and .META. exist as HBase tables. They are filtered out
of the HBase shell's <code>list</code> command, but they are in fact tables just like any other.
</para>
<section xml:id="arch.catalog.root">
<title>ROOT</title>
<para>-ROOT- keeps track of where the .META. table is. The -ROOT- table structure is as follows:
</para>
<para>Key:
<itemizedlist>
<listitem>.META. region key (<code>.META.,,1</code>)</listitem>
</itemizedlist>
</para>
<para>Values:
<itemizedlist>
<listitem><code>info:regioninfo</code> (serialized <link xlink:href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/HRegionInfo.html">HRegionInfo</link>
instance of .META.)</listitem>
<listitem><code>info:server</code> (server:port of the RegionServer holding .META.)</listitem>
<listitem><code>info:serverstartcode</code> (start-time of the RegionServer process holding .META.)</listitem>
</itemizedlist>
</para>
</section>
<section xml:id="arch.catalog.meta">
<title>META</title>
<para>The .META. table keeps a list of all regions in the system. The .META. table structure is as follows:
</para>
<para>Key:
<itemizedlist>
<listitem>Region key of the format (<code>[table],[region start key],[region id]</code>)</listitem>
</itemizedlist>
</para>
<para>Values:
<itemizedlist>
<listitem><code>info:regioninfo</code> (serialized <link xlink:href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/HRegionInfo.html">
HRegionInfo</link> instance for this region)
</listitem>
<listitem><code>info:server</code> (server:port of the RegionServer containing this region)</listitem>
<listitem><code>info:serverstartcode</code> (start-time of the RegionServer process containing this region)</listitem>
</itemizedlist>
</para>
<para>When a table is in the process of splitting two other columns will be created, <code>info:splitA</code> and <code>info:splitB</code>
which represent the two daughter regions. The values for these columns are also serialized HRegionInfo instances.
After the region has been split eventually this row will be deleted.
</para>
<para>Notes on HRegionInfo: the empty key is used to denote table start and table end. A region with an empty start key
is the first region in a table. If region has both an empty start and an empty end key, it's the only region in the table
</para>
<para>In the (hopefully unlikely) event that programmatic processing of catalog metadata is required, see the
<link xlink:href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/util/Writables.html#getHRegionInfo%28byte[]%29">Writables</link> utility.
</para>
</section>
<section xml:id="arch.catalog.startup">
<title>Startup Sequencing</title>
<para>The META location is set in ROOT first. Then META is updated with server and startcode values.
</para>
<para>For information on region-RegionServer assignment, see <xref linkend="regions.arch.assignment"/>.
</para>
</section>
</section> <!-- catalog -->
<section xml:id="client">
<title>Client</title>
<para>The HBase client
<link xlink:href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/client/HTable.html">HTable</link>
is responsible for finding RegionServers that are serving the
particular row range of interest. It does this by querying
the <code>.META.</code> and <code>-ROOT-</code> catalog tables
(TODO: Explain). After locating the required
region(s), the client <emphasis>directly</emphasis> contacts
the RegionServer serving that region (i.e., it does not go
through the master) and issues the read or write request.
This information is cached in the client so that subsequent requests
need not go through the lookup process. Should a region be reassigned
either by the master load balancer or because a RegionServer has died,
the client will requery the catalog tables to determine the new
location of the user region.
</para>
<para>See <xref linkend="master.runtime"/> for more information about the impact of the Master on HBase Client
communication.
</para>
<para>Administrative functions are handled through <link xlink:href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/client/HBaseAdmin.html">HBaseAdmin</link>
</para>
<section xml:id="client.connections"><title>Connections</title>
<para>For connection configuration information, see <xref linkend="client_dependencies" />.
</para>
<para><emphasis><link xlink:href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/client/HTable.html">HTable</link>
instances are not thread-safe</emphasis>. Only one thread use an instance of HTable at any given
time. When creating HTable instances, it is advisable to use the same <link xlink:href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/HBaseConfiguration">HBaseConfiguration</link>
instance. This will ensure sharing of ZooKeeper and socket instances to the RegionServers
which is usually what you want. For example, this is preferred:
<programlisting>HBaseConfiguration conf = HBaseConfiguration.create();
HTable table1 = new HTable(conf, "myTable");
HTable table2 = new HTable(conf, "myTable");</programlisting>
as opposed to this:
<programlisting>HBaseConfiguration conf1 = HBaseConfiguration.create();
HTable table1 = new HTable(conf1, "myTable");
HBaseConfiguration conf2 = HBaseConfiguration.create();
HTable table2 = new HTable(conf2, "myTable");</programlisting>
For more information about how connections are handled in the HBase client,
see <link xlink:href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/client/HConnectionManager.html">HConnectionManager</link>.
</para>
<section xml:id="client.connection.pooling"><title>Connection Pooling</title>
<para>For applications which require high-end multithreaded access (e.g., web-servers or application servers that may serve many application threads
in a single JVM), one solution is <link xlink:href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/client/HTablePool.html">HTablePool</link>.
But as written currently, it is difficult to control client resource consumption when using HTablePool.
</para>
<para>
Another solution is to precreate an <classname>HConnection</classname> using
<programlisting>// Create a connection to the cluster.
HConnection connection = HConnectionManager.createConnection(Configuration);
HTableInterface table = connection.getTable("myTable");
// use table as needed, the table returned is lightweight
table.close();
// use the connection for other access to the cluster
connection.close();</programlisting>
Constructing HTableInterface implementation is very lightweight and resources are controlled/shared if you go this route.
</para>
</section>
</section>
<section xml:id="client.writebuffer"><title>WriteBuffer and Batch Methods</title>
<para>If <xref linkend="perf.hbase.client.autoflush" /> is turned off on
<link xlink:href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/client/HTable.html">HTable</link>,
<classname>Put</classname>s are sent to RegionServers when the writebuffer
is filled. The writebuffer is 2MB by default. Before an HTable instance is
discarded, either <methodname>close()</methodname> or
<methodname>flushCommits()</methodname> should be invoked so Puts
will not be lost.
</para>
<para>Note: <code>htable.delete(Delete);</code> does not go in the writebuffer! This only applies to Puts.
</para>
<para>For additional information on write durability, review the <link xlink:href="../acid-semantics.html">ACID semantics</link> page.
</para>
<para>For fine-grained control of batching of
<classname>Put</classname>s or <classname>Delete</classname>s,
see the <link xlink:href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/client/HTable.html#batch%28java.util.List%29">batch</link> methods on HTable.
</para>
</section>
<section xml:id="client.external"><title>External Clients</title>
<para>Information on non-Java clients and custom protocols is covered in <xref linkend="external_apis" />
</para>
</section>
<section xml:id="client.rowlocks"><title>RowLocks</title>
<para><link xlink:href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/client/HTable.html#lockRow%28byte[]%29">RowLocks</link> are still
in the client API <emphasis>however</emphasis> they are discouraged because if not managed properly these can
lock up the RegionServers.
</para>
<para>There is an oustanding ticket <link xlink:href="https://issues.apache.org/jira/browse/HBASE-2332">HBASE-2332</link> to
remove this feature from the client.
</para>
</section>
</section>
<section xml:id="client.filter"><title>Client Request Filters</title>
<para><link xlink:href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/client/Get.html">Get</link> and <link xlink:href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/client/Scan.html">Scan</link> instances can be
optionally configured with <link xlink:href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/filter/Filter.html">filters</link> which are applied on the RegionServer.
</para>
<para>Filters can be confusing because there are many different types, and it is best to approach them by understanding the groups
of Filter functionality.
</para>
<section xml:id="client.filter.structural"><title>Structural</title>
<para>Structural Filters contain other Filters.</para>
<section xml:id="client.filter.structural.fl"><title>FilterList</title>
<para><link xlink:href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/filter/FilterList.html">FilterList</link>
represents a list of Filters with a relationship of <code>FilterList.Operator.MUST_PASS_ALL</code> or
<code>FilterList.Operator.MUST_PASS_ONE</code> between the Filters. The following example shows an 'or' between two
Filters (checking for either 'my value' or 'my other value' on the same attribute).
<programlisting>
FilterList list = new FilterList(FilterList.Operator.MUST_PASS_ONE);
SingleColumnValueFilter filter1 = new SingleColumnValueFilter(
cf,
column,
CompareOp.EQUAL,
Bytes.toBytes("my value")
);
list.add(filter1);
SingleColumnValueFilter filter2 = new SingleColumnValueFilter(
cf,
column,
CompareOp.EQUAL,
Bytes.toBytes("my other value")
);
list.add(filter2);
scan.setFilter(list);
</programlisting>
</para>
</section>
</section>
<section xml:id="client.filter.cv"><title>Column Value</title>
<section xml:id="client.filter.cv.scvf"><title>SingleColumnValueFilter</title>
<para><link xlink:href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/filter/SingleColumnValueFilter.html">SingleColumnValueFilter</link>
can be used to test column values for equivalence (<code><link xlink:href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/filter/CompareFilter.CompareOp.html">CompareOp.EQUAL</link>
</code>), inequality (<code>CompareOp.NOT_EQUAL</code>), or ranges
(e.g., <code>CompareOp.GREATER</code>). The folowing is example of testing equivalence a column to a String value "my value"...
<programlisting>
SingleColumnValueFilter filter = new SingleColumnValueFilter(
cf,
column,
CompareOp.EQUAL,
Bytes.toBytes("my value")
);
scan.setFilter(filter);
</programlisting>
</para>
</section>
</section>
<section xml:id="client.filter.cvp"><title>Column Value Comparators</title>
<para>There are several Comparator classes in the Filter package that deserve special mention.
These Comparators are used in concert with other Filters, such as <xref linkend="client.filter.cv.scvf" />.
</para>
<section xml:id="client.filter.cvp.rcs"><title>RegexStringComparator</title>
<para><link xlink:href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/filter/RegexStringComparator.html">RegexStringComparator</link>
supports regular expressions for value comparisons.
<programlisting>
RegexStringComparator comp = new RegexStringComparator("my."); // any value that starts with 'my'
SingleColumnValueFilter filter = new SingleColumnValueFilter(
cf,
column,
CompareOp.EQUAL,
comp
);
scan.setFilter(filter);
</programlisting>
See the Oracle JavaDoc for <link xlink:href="http://download.oracle.com/javase/6/docs/api/java/util/regex/Pattern.html">supported RegEx patterns in Java</link>.
</para>
</section>
<section xml:id="client.filter.cvp.rcs"><title>SubstringComparator</title>
<para><link xlink:href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/filter/SubstringComparator.html">SubstringComparator</link>
can be used to determine if a given substring exists in a value. The comparison is case-insensitive.
</para>
<programlisting>
SubstringComparator comp = new SubstringComparator("y val"); // looking for 'my value'
SingleColumnValueFilter filter = new SingleColumnValueFilter(
cf,
column,
CompareOp.EQUAL,
comp
);
scan.setFilter(filter);
</programlisting>
</section>
<section xml:id="client.filter.cvp.bfp"><title>BinaryPrefixComparator</title>
<para>See <link xlink:href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/filter/BinaryPrefixComparator.html">BinaryPrefixComparator</link>.</para>
</section>
<section xml:id="client.filter.cvp.bc"><title>BinaryComparator</title>
<para>See <link xlink:href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/filter/BinaryComparator.html">BinaryComparator</link>.</para>
</section>
</section>
<section xml:id="client.filter.kvm"><title>KeyValue Metadata</title>
<para>As HBase stores data internally as KeyValue pairs, KeyValue Metadata Filters evaluate the existence of keys (i.e., ColumnFamily:Column qualifiers)
for a row, as opposed to values the previous section.
</para>
<section xml:id="client.filter.kvm.ff"><title>FamilyFilter</title>
<para><link xlink:href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/filter/FamilyFilter.html">FamilyFilter</link> can be used
to filter on the ColumnFamily. It is generally a better idea to select ColumnFamilies in the Scan than to do it with a Filter.</para>
</section>
<section xml:id="client.filter.kvm.qf"><title>QualifierFilter</title>
<para><link xlink:href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/filter/QualifierFilter.html">QualifierFilter</link> can be used
to filter based on Column (aka Qualifier) name.
</para>
</section>
<section xml:id="client.filter.kvm.cpf"><title>ColumnPrefixFilter</title>
<para><link xlink:href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/filter/ColumnPrefixFilter.html">ColumnPrefixFilter</link> can be used
to filter based on the lead portion of Column (aka Qualifier) names.
</para>
<para>A ColumnPrefixFilter seeks ahead to the first column matching the prefix in each row and for each involved column family. It can be used to efficiently
get a subset of the columns in very wide rows.
</para>
<para>Note: The same column qualifier can be used in different column families. This filter returns all matching columns.
</para>
<para>Example: Find all columns in a row and family that start with "abc"
<programlisting>
HTableInterface t = ...;
byte[] row = ...;
byte[] family = ...;
byte[] prefix = Bytes.toBytes("abc");
Scan scan = new Scan(row, row); // (optional) limit to one row
scan.addFamily(family); // (optional) limit to one family
Filter f = new ColumnPrefixFilter(prefix);
scan.setFilter(f);
scan.setBatch(10); // set this if there could be many columns returned
ResultScanner rs = t.getScanner(scan);
for (Result r = rs.next(); r != null; r = rs.next()) {
for (KeyValue kv : r.raw()) {
// each kv represents a column
}
}
rs.close();
</programlisting>
</para>
</section>
<section xml:id="client.filter.kvm.mcpf"><title>MultipleColumnPrefixFilter</title>
<para><link xlink:href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/filter/MultipleColumnPrefixFilter.html">MultipleColumnPrefixFilter</link> behaves like ColumnPrefixFilter
but allows specifying multiple prefixes.
</para>
<para>Like ColumnPrefixFilter, MultipleColumnPrefixFilter efficiently seeks ahead to the first column matching the lowest prefix and also seeks past ranges of columns between prefixes.
It can be used to efficiently get discontinuous sets of columns from very wide rows.
</para>
<para>Example: Find all columns in a row and family that start with "abc" or "xyz"
<programlisting>
HTableInterface t = ...;
byte[] row = ...;
byte[] family = ...;
byte[][] prefixes = new byte[][] {Bytes.toBytes("abc"), Bytes.toBytes("xyz")};
Scan scan = new Scan(row, row); // (optional) limit to one row
scan.addFamily(family); // (optional) limit to one family
Filter f = new MultipleColumnPrefixFilter(prefixes);
scan.setFilter(f);
scan.setBatch(10); // set this if there could be many columns returned
ResultScanner rs = t.getScanner(scan);
for (Result r = rs.next(); r != null; r = rs.next()) {
for (KeyValue kv : r.raw()) {
// each kv represents a column
}
}
rs.close();
</programlisting>
</para>
</section>
<section xml:id="client.filter.kvm.crf "><title>ColumnRangeFilter</title>
<para>A <link xlink:href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/filter/ColumnRangeFilter.html">ColumnRangeFilter</link> allows efficient intra row scanning.
</para>
<para>A ColumnRangeFilter can seek ahead to the first matching column for each involved column family. It can be used to efficiently
get a 'slice' of the columns of a very wide row.
i.e. you have a million columns in a row but you only want to look at columns bbbb-bbdd.
</para>
<para>Note: The same column qualifier can be used in different column families. This filter returns all matching columns.
</para>
<para>Example: Find all columns in a row and family between "bbbb" (inclusive) and "bbdd" (inclusive)
<programlisting>
HTableInterface t = ...;
byte[] row = ...;
byte[] family = ...;
byte[] startColumn = Bytes.toBytes("bbbb");
byte[] endColumn = Bytes.toBytes("bbdd");
Scan scan = new Scan(row, row); // (optional) limit to one row
scan.addFamily(family); // (optional) limit to one family
Filter f = new ColumnRangeFilter(startColumn, true, endColumn, true);
scan.setFilter(f);
scan.setBatch(10); // set this if there could be many columns returned
ResultScanner rs = t.getScanner(scan);
for (Result r = rs.next(); r != null; r = rs.next()) {
for (KeyValue kv : r.raw()) {
// each kv represents a column
}
}
rs.close();
</programlisting>
</para>
<para>Note: Introduced in HBase 0.92</para>
</section>
</section>
<section xml:id="client.filter.row"><title>RowKey</title>
<section xml:id="client.filter.row.rf"><title>RowFilter</title>
<para>It is generally a better idea to use the startRow/stopRow methods on Scan for row selection, however
<link xlink:href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/filter/RowFilter.html">RowFilter</link> can also be used.</para>
</section>
</section>
<section xml:id="client.filter.utility"><title>Utility</title>
<section xml:id="client.filter.utility.fkof"><title>FirstKeyOnlyFilter</title>
<para>This is primarily used for rowcount jobs.
See <link xlink:href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/filter/FirstKeyOnlyFilter.html">FirstKeyOnlyFilter</link>.</para>
</section>
</section>
</section> <!-- client.filter -->
<section xml:id="master"><title>Master</title>
<para><code>HMaster</code> is the implementation of the Master Server. The Master server
is responsible for monitoring all RegionServer instances in the cluster, and is
the interface for all metadata changes. In a distributed cluster, the Master typically runs on the <xref linkend="arch.hdfs.nn" /><footnote>
<para>J Mohamed Zahoor goes into some more detail on the Master Architecture in this blog posting, <link
xlink:href="http://blog.zahoor.in/2012/08/hbase-hmaster-architecture/">HBase HMaster Architecture
</link>.</para>
</footnote>
</para>
<section xml:id="master.startup"><title>Startup Behavior</title>
<para>If run in a multi-Master environment, all Masters compete to run the cluster. If the active
Master loses its lease in ZooKeeper (or the Master shuts down), then then the remaining Masters jostle to
take over the Master role.
</para>
</section>
<section xml:id="master.runtime"><title>Runtime Impact</title>
<para>A common dist-list question is what happens to an HBase cluster when the Master goes down. Because the
HBase client talks directly to the RegionServers, the cluster can still function in a "steady
state." Additionally, per <xref linkend="arch.catalog"/> ROOT and META exist as HBase tables (i.e., are
not resident in the Master). However, the Master controls critical functions such as RegionServer failover and
completing region splits. So while the cluster can still run <emphasis>for a time</emphasis> without the Master,
the Master should be restarted as soon as possible.
</para>
</section>
<section xml:id="master.api"><title>Interface</title>
<para>The methods exposed by <code>HMasterInterface</code> are primarily metadata-oriented methods:
<itemizedlist>
<listitem>Table (createTable, modifyTable, removeTable, enable, disable)
</listitem>
<listitem>ColumnFamily (addColumn, modifyColumn, removeColumn)
</listitem>
<listitem>Region (move, assign, unassign)
</listitem>
</itemizedlist>
For example, when the <code>HBaseAdmin</code> method <code>disableTable</code> is invoked, it is serviced by the Master server.
</para>
</section>
<section xml:id="master.processes"><title>Processes</title>
<para>The Master runs several background threads:
</para>
<section xml:id="master.processes.loadbalancer"><title>LoadBalancer</title>
<para>Periodically, and when there are no regions in transition,
a load balancer will run and move regions around to balance the cluster's load.
See <xref linkend="balancer_config" /> for configuring this property.</para>
<para>See <xref linkend="regions.arch.assignment"/> for more information on region assignment.
</para>
</section>
<section xml:id="master.processes.catalog"><title>CatalogJanitor</title>
<para>Periodically checks and cleans up the .META. table. See <xref linkend="arch.catalog.meta" /> for more information on META.</para>
</section>
</section>
</section>
<section xml:id="regionserver.arch"><title>RegionServer</title>
<para><code>HRegionServer</code> is the RegionServer implementation. It is responsible for serving and managing regions.
In a distributed cluster, a RegionServer runs on a <xref linkend="arch.hdfs.dn" />.
</para>
<section xml:id="regionserver.arch.api"><title>Interface</title>
<para>The methods exposed by <code>HRegionRegionInterface</code> contain both data-oriented and region-maintenance methods:
<itemizedlist>
<listitem>Data (get, put, delete, next, etc.)
</listitem>
<listitem>Region (splitRegion, compactRegion, etc.)
</listitem>
</itemizedlist>
For example, when the <code>HBaseAdmin</code> method <code>majorCompact</code> is invoked on a table, the client is actually iterating through
all regions for the specified table and requesting a major compaction directly to each region.
</para>
</section>
<section xml:id="regionserver.arch.processes"><title>Processes</title>
<para>The RegionServer runs a variety of background threads:</para>
<section xml:id="regionserver.arch.processes.compactsplit"><title>CompactSplitThread</title>
<para>Checks for splits and handle minor compactions.</para>
</section>
<section xml:id="regionserver.arch.processes.majorcompact"><title>MajorCompactionChecker</title>
<para>Checks for major compactions.</para>
</section>
<section xml:id="regionserver.arch.processes.memstore"><title>MemStoreFlusher</title>
<para>Periodically flushes in-memory writes in the MemStore to StoreFiles.</para>
</section>
<section xml:id="regionserver.arch.processes.log"><title>LogRoller</title>
<para>Periodically checks the RegionServer's HLog.</para>
</section>
</section>
<section xml:id="coprocessors"><title>Coprocessors</title>
<para>Coprocessors were added in 0.92. There is a thorough <link xlink:href="https://blogs.apache.org/hbase/entry/coprocessor_introduction">Blog Overview of CoProcessors</link>
posted. Documentation will eventually move to this reference guide, but the blog is the most current information available at this time.
</para>
</section>
<section xml:id="block.cache">
<title>Block Cache</title>
<section xml:id="block.cache.design">
<title>Design</title>
<para>The Block Cache is an LRU cache that contains three levels of block priority to allow for scan-resistance and in-memory ColumnFamilies:
</para>
<itemizedlist>
<listitem>Single access priority: The first time a block is loaded from HDFS it normally has this priority and it will be part of the first group to be considered
during evictions. The advantage is that scanned blocks are more likely to get evicted than blocks that are getting more usage.
</listitem>
<listitem>Mutli access priority: If a block in the previous priority group is accessed again, it upgrades to this priority. It is thus part of the second group
considered during evictions.
</listitem>
<listitem>In-memory access priority: If the block's family was configured to be "in-memory", it will be part of this priority disregarding the number of times it
was accessed. Catalog tables are configured like this. This group is the last one considered during evictions.
</listitem>
</itemizedlist>
<para>
For more information, see the <link xlink:href="http://hbase.apache.org/xref/org/apache/hadoop/hbase/io/hfile/LruBlockCache.html">LruBlockCache source</link>
</para>
</section>
<section xml:id="block.cache.usage">
<title>Usage</title>
<para>Block caching is enabled by default for all the user tables which means that any read operation will load the LRU cache. This might be good for a large number of use cases,
but further tunings are usually required in order to achieve better performance. An important concept is the
<link xlink:href="http://en.wikipedia.org/wiki/Working_set_size">working set size</link>, or WSS, which is: "the amount of memory needed to compute the answer to a problem".
For a website, this would be the data that's needed to answer the queries over a short amount of time.
</para>
<para>The way to calculate how much memory is available in HBase for caching is:
</para>
<programlisting>
number of region servers * heap size * hfile.block.cache.size * 0.85
</programlisting>
<para>The default value for the block cache is 0.25 which represents 25% of the available heap. The last value (85%) is the default acceptable loading factor in the LRU cache after
which eviction is started. The reason it is included in this equation is that it would be unrealistic to say that it is possible to use 100% of the available memory since this would
make the process blocking from the point where it loads new blocks. Here are some examples:
</para>
<itemizedlist>
<listitem>One region server with the default heap size (1GB) and the default block cache size will have 217MB of block cache available.
</listitem>
<listitem>20 region servers with the heap size set to 8GB and a default block cache size will have 34GB of block cache.
</listitem>
<listitem>100 region servers with the heap size set to 24GB and a block cache size of 0.5 will have about 1TB of block cache.
</listitem>
</itemizedlist>
<para>Your data isn't the only resident of the block cache, here are others that you may have to take into account:
</para>
<itemizedlist>
<listitem>Catalog tables: The -ROOT- and .META. tables are forced into the block cache and have the in-memory priority which means that they are harder to evict. The former never uses
more than a few hundreds of bytes while the latter can occupy a few MBs (depending on the number of regions).
</listitem>
<listitem>HFiles indexes: HFile is the file format that HBase uses to store data in HDFS and it contains a multi-layered index in order seek to the data without having to read the whole file.
The size of those indexes is a factor of the block size (64KB by default), the size of your keys and the amount of data you are storing. For big data sets it's not unusual to see numbers around
1GB per region server, although not all of it will be in cache because the LRU will evict indexes that aren't used.
</listitem>
<listitem>Keys: Taking into account only the values that are being stored is missing half the picture since every value is stored along with its keys
(row key, family, qualifier, and timestamp). See <xref linkend="keysize"/>.
</listitem>
<listitem>Bloom filters: Just like the HFile indexes, those data structures (when enabled) are stored in the LRU.
</listitem>
</itemizedlist>
<para>Currently the recommended way to measure HFile indexes and bloom filters sizes is to look at the region server web UI and checkout the relevant metrics. For keys,
sampling can be done by using the HFile command line tool and look for the average key size metric.
</para>
<para>It's generally bad to use block caching when the WSS doesn't fit in memory. This is the case when you have for example 40GB available across all your region servers' block caches
but you need to process 1TB of data. One of the reasons is that the churn generated by the evictions will trigger more garbage collections unnecessarily. Here are two use cases:
</para>
<itemizedlist>
<listitem>Fully random reading pattern: This is a case where you almost never access the same row twice within a short amount of time such that the chance of hitting a cached block is close
to 0. Setting block caching on such a table is a waste of memory and CPU cycles, more so that it will generate more garbage to pick up by the JVM. For more information on monitoring GC,
see <xref linkend="trouble.log.gc"/>.
</listitem>
<listitem>Mapping a table: In a typical MapReduce job that takes a table in input, every row will be read only once so there's no need to put them into the block cache. The Scan object has
the option of turning this off via the setCaching method (set it to false). You can still keep block caching turned on on this table if you need fast random read access. An example would be
counting the number of rows in a table that serves live traffic, caching every block of that table would create massive churn and would surely evict data that's currently in use.
</listitem>
</itemizedlist>
</section>
</section>
<section xml:id="wal">
<title >Write Ahead Log (WAL)</title>
<section xml:id="purpose.wal">
<title>Purpose</title>
<para>Each RegionServer adds updates (Puts, Deletes) to its write-ahead log (WAL)
first, and then to the <xref linkend="store.memstore"/> for the affected <xref linkend="store" />.
This ensures that HBase has durable writes. Without WAL, there is the possibility of data loss in the case of a RegionServer failure
before each MemStore is flushed and new StoreFiles are written. <link xlink:href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/regionserver/wal/HLog.html">HLog</link>
is the HBase WAL implementation, and there is one HLog instance per RegionServer.
</para>The WAL is in HDFS in <filename>/hbase/.logs/</filename> with subdirectories per region.
<para>
For more general information about the concept of write ahead logs, see the Wikipedia
<link xlink:href="http://en.wikipedia.org/wiki/Write-ahead_logging">Write-Ahead Log</link> article.
</para>
</section>
<section xml:id="wal_flush">
<title>WAL Flushing</title>
<para>TODO (describe).
</para>
</section>
<section xml:id="wal_splitting">
<title>WAL Splitting</title>
<section><title>How edits are recovered from a crashed RegionServer</title>
<para>When a RegionServer crashes, it will lose its ephemeral lease in
ZooKeeper...TODO</para>
</section>
<section>
<title><varname>hbase.hlog.split.skip.errors</varname></title>
<para>When set to <constant>true</constant>, any error
encountered splitting will be logged, the problematic WAL will be
moved into the <filename>.corrupt</filename> directory under the hbase
<varname>rootdir</varname>, and processing will continue. If set to
<constant>false</constant>, the default, the exception will be propagated and the
split logged as failed.<footnote>
<para>See <link
xlink:href="https://issues.apache.org/jira/browse/HBASE-2958">HBASE-2958
When hbase.hlog.split.skip.errors is set to false, we fail the
split but thats it</link>. We need to do more than just fail split
if this flag is set.</para>
</footnote></para>
</section>
<section>
<title>How EOFExceptions are treated when splitting a crashed
RegionServers' WALs</title>
<para>If we get an EOF while splitting logs, we proceed with the split
even when <varname>hbase.hlog.split.skip.errors</varname> ==
<constant>false</constant>. An EOF while reading the last log in the
set of files to split is near-guaranteed since the RegionServer likely
crashed mid-write of a record. But we'll continue even if we got an
EOF reading other than the last file in the set.<footnote>
<para>For background, see <link
xlink:href="https://issues.apache.org/jira/browse/HBASE-2643">HBASE-2643
Figure how to deal with eof splitting logs</link></para>
</footnote></para>
</section>
</section>
</section>
</section> <!-- regionserver -->
<section xml:id="regions.arch">
<title>Regions</title>
<para>Regions are the basic element of availability and
distribution for tables, and are comprised of a Store per Column Family. The heirarchy of objects
is as follows:
<programlisting>
<filename>Table</filename> (HBase table)
<filename>Region</filename> (Regions for the table)
<filename>Store</filename> (Store per ColumnFamily for each Region for the table)
<filename>MemStore</filename> (MemStore for each Store for each Region for the table)
<filename>StoreFile</filename> (StoreFiles for each Store for each Region for the table)
<filename>Block</filename> (Blocks within a StoreFile within a Store for each Region for the table)
</programlisting>
For a description of what HBase files look like when written to HDFS, see <xref linkend="trouble.namenode.hbase.objects"/>.
</para>
<section xml:id="arch.regions.size">
<para> In general, HBase is designed to run with a small (20-200) number of relatively large (5-20Gb) regions per server. The considerations for this are as follows:</para>
<section xml:id="too_many_regions">
<title>Why cannot I have too many regions?</title>
<para>
Typically you want to keep your region count low on HBase for numerous reasons.
Usually right around 100 regions per RegionServer has yielded the best results.
Here are some of the reasons below for keeping region count low:
<orderedlist>
<listitem><para>
MSLAB requires 2mb per memstore (that's 2mb per family per region).
1000 regions that have 2 families each is 3.9GB of heap used, and it's not even storing data yet. NB: the 2MB value is configurable.
</para></listitem>
<listitem><para>If you fill all the regions at somewhat the same rate, the global memory usage makes it that it forces tiny
flushes when you have too many regions which in turn generates compactions.
Rewriting the same data tens of times is the last thing you want.
An example is filling 1000 regions (with one family) equally and let's consider a lower bound for global memstore
usage of 5GB (the region server would have a big heap).
Once it reaches 5GB it will force flush the biggest region,
at that point they should almost all have about 5MB of data so
it would flush that amount. 5MB inserted later, it would flush another
region that will now have a bit over 5MB of data, and so on.
This is currently the main limiting factor for the number of regions; see <xref linkend="ops.capacity.regions.count" />
for detailed formula.
</para></listitem>
<listitem><para>The master as is is allergic to tons of regions, and will
take a lot of time assigning them and moving them around in batches.
The reason is that it's heavy on ZK usage, and it's not very async
at the moment (could really be improved -- and has been imporoved a bunch
in 0.96 hbase).
</para></listitem>
<listitem><para>
In older versions of HBase (pre-v2 hfile, 0.90 and previous), tons of regions
on a few RS can cause the store file index to rise, increasing heap usage and potentially
creating memory pressure or OOME on the RSs
</para></listitem>
</orderedlist>
</para>
</section>
<para>Another issue is the effect of the number of regions on mapreduce jobs; it is typical to have one mapper per HBase region.
Thus, hosting only 5 regions per RS may not be enough to get sufficient number of tasks for a mapreduce job, while 1000 regions will generate far too many tasks.
</para>
<para>See <xref linkend="ops.capacity.regions" /> for configuration guidelines.</para>
</section>
<section xml:id="regions.arch.assignment">
<title>Region-RegionServer Assignment</title>
<para>This section describes how Regions are assigned to RegionServers.
</para>
<section xml:id="regions.arch.assignment.startup">
<title>Startup</title>
<para>When HBase starts regions are assigned as follows (short version):
<orderedlist>
<listitem>The Master invokes the <code>AssignmentManager</code> upon startup.
</listitem>
<listitem>The <code>AssignmentManager</code> looks at the existing region assignments in META.
</listitem>
<listitem>If the region assignment is still valid (i.e., if the RegionServer is still online)
then the assignment is kept.
</listitem>
<listitem>If the assignment is invalid, then the <code>LoadBalancerFactory</code> is invoked to assign the
region. The <code>DefaultLoadBalancer</code> will randomly assign the region to a RegionServer.
</listitem>
<listitem>META is updated with the RegionServer assignment (if needed) and the RegionServer start codes
(start time of the RegionServer process) upon region opening by the RegionServer.
</listitem>
</orderedlist>
</para>
</section>
<section xml:id="regions.arch.assignment.failover">
<title>Failover</title>
<para>When a RegionServer fails (short version):
<orderedlist>
<listitem>The regions immediately become unavailable because the RegionServer is down.
</listitem>
<listitem>The Master will detect that the RegionServer has failed.
</listitem>
<listitem>The region assignments will be considered invalid and will be re-assigned just
like the startup sequence.
</listitem>
</orderedlist>
</para>
</section>
<section xml:id="regions.arch.balancer">
<title>Region Load Balancing</title>
<para>
Regions can be periodically moved by the <xref linkend="master.processes.loadbalancer" />.
</para>
</section>
</section> <!-- assignment -->
<section xml:id="regions.arch.locality">
<title>Region-RegionServer Locality</title>
<para>Over time, Region-RegionServer locality is achieved via HDFS block replication.
The HDFS client does the following by default when choosing locations to write replicas:
<orderedlist>
<listitem>First replica is written to local node
</listitem>
<listitem>Second replica is written to a random node on another rack
</listitem>
<listitem>Third replica is written on the same rack as the second, but on a different node chosen randomly
</listitem>
<listitem>Subsequent replicas are written on random nodes on the cluster
<footnote><para>See <emphasis>Replica Placement: The First Baby Steps</emphasis> on this page: <link xlink:href="http://hadoop.apache.org/docs/stable/hadoop-project-dist/hadoop-hdfs/HdfsDesign.html">HDFS Architecture</link></para></footnote>
</listitem>
</orderedlist>
Thus, HBase eventually achieves locality for a region after a flush or a compaction.
In a RegionServer failover situation a RegionServer may be assigned regions with non-local
StoreFiles (because none of the replicas are local), however as new data is written
in the region, or the table is compacted and StoreFiles are re-written, they will become "local"
to the RegionServer.
</para>
<para>For more information, see <emphasis>Replica Placement: The First Baby Steps</emphasis> on this page: <link xlink:href="http://hadoop.apache.org/docs/stable/hadoop-project-dist/hadoop-hdfs/HdfsDesign.html">HDFS Architecture</link>
and also Lars George's blog on <link xlink:href="http://www.larsgeorge.com/2010/05/hbase-file-locality-in-hdfs.html">HBase and HDFS locality</link>.
</para>
</section>
<section xml:id="arch.region.splits">
<title>Region Splits</title>
<para>Regions split when they reach a configured threshold.
Below we treat the topic in short. For a longer exposition,
see <link xlink:href="http://hortonworks.com/blog/apache-hbase-region-splitting-and-merging/">Apache HBase Region Splitting and Merging</link>
by our Enis Soztutar.
</para>
<para>Splits run unaided on the RegionServer; i.e. the Master does not
participate. The RegionServer splits a region, offlines the split
region and then adds the daughter regions to META, opens daughters on
the parent's hosting RegionServer and then reports the split to the
Master. See <xref linkend="disable.splitting" /> for how to manually manage
splits (and for why you might do this)</para>
<section>
<title>Custom Split Policies</title>
<para>The default split policy can be overwritten using a custom <link xlink:href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/regionserver/RegionSplitPolicy.html">RegionSplitPolicy</link> (HBase 0.94+).
Typically a custom split policy should extend HBase's default split policy: <link xlink:href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/regionserver/ConstantSizeRegionSplitPolicy.html">ConstantSizeRegionSplitPolicy</link>.
</para>
<para>The policy can set globally through the HBaseConfiguration used or on a per table basis:
<programlisting>
HTableDescriptor myHtd = ...;
myHtd.setValue(HTableDescriptor.SPLIT_POLICY, MyCustomSplitPolicy.class.getName());
</programlisting>
</para>
</section>
</section>
<section>
<title>Online Region Merges</title>
<para>Both Master and Regionserver participate in the event of online region merges.
Client sends merge RPC to master, then master moves the regions together to the
same regionserver where the more heavily loaded region resided, finally master
send merge request to this regionserver and regionserver run the region merges.
Similar with process of region splits, region merges run as a local transaction
on the regionserver, offlines the regions and then merges two regions on the file
system, atomically delete merging regions from META and add merged region to the META,
opens merged region on the regionserver and reports the merge to Master at last.
</para>
<para>An example of region merges in the hbase shell
<programlisting>$ hbase> merge_region 'ENCODED_REGIONNAME', 'ENCODED_REGIONNAME'
hbase> merge_region 'ENCODED_REGIONNAME', 'ENCODED_REGIONNAME', true
</programlisting>
It's an asynchronous operation and call returns immediately without waiting merge completed.
Passing 'true' as the optional third parameter will force a merge ('force' merges regardless
else merge will fail unless passed adjacent regions. 'force' is for expert use only)
</para>
</section>
<section xml:id="store">
<title>Store</title>
<para>A Store hosts a MemStore and 0 or more StoreFiles (HFiles). A Store corresponds to a column family for a table for a given region.
</para>
<section xml:id="store.memstore">
<title>MemStore</title>
<para>The MemStore holds in-memory modifications to the Store. Modifications are KeyValues.
When asked to flush, current memstore is moved to snapshot and is cleared.
HBase continues to serve edits out of new memstore and backing snapshot until flusher reports in that the
flush succeeded. At this point the snapshot is let go.</para>
</section>
<section xml:id="hfile">
<title>StoreFile (HFile)</title>
<para>StoreFiles are where your data lives.
</para>
<section><title>HFile Format</title>
<para>The <emphasis>hfile</emphasis> file format is based on
the SSTable file described in the <link xlink:href="http://research.google.com/archive/bigtable.html">BigTable [2006]</link> paper and on
Hadoop's <link xlink:href="http://hadoop.apache.org/common/docs/current/api/org/apache/hadoop/io/file/tfile/TFile.html">tfile</link>
(The unit test suite and the compression harness were taken directly from tfile).
Schubert Zhang's blog post on <link xlink:ref="http://cloudepr.blogspot.com/2009/09/hfile-block-indexed-file-format-to.html">HFile: A Block-Indexed File Format to Store Sorted Key-Value Pairs</link> makes for a thorough introduction to HBase's hfile. Matteo Bertozzi has also put up a
helpful description, <link xlink:href="http://th30z.blogspot.com/2011/02/hbase-io-hfile.html?spref=tw">HBase I/O: HFile</link>.
</para>
<para>For more information, see the <link xlink:href="http://hbase.apache.org/xref/org/apache/hadoop/hbase/io/hfile/HFile.html">HFile source code</link>.
Also see <xref linkend="hfilev2"/> for information about the HFile v2 format that was included in 0.92.
</para>
</section>
<section xml:id="hfile_tool">
<title>HFile Tool</title>
<para>To view a textualized version of hfile content, you can do use
the <classname>org.apache.hadoop.hbase.io.hfile.HFile
</classname>tool. Type the following to see usage:<programlisting><code>$ ${HBASE_HOME}/bin/hbase org.apache.hadoop.hbase.io.hfile.HFile </code> </programlisting>For
example, to view the content of the file
<filename>hdfs://10.81.47.41:8020/hbase/TEST/1418428042/DSMP/4759508618286845475</filename>,
type the following:<programlisting> <code>$ ${HBASE_HOME}/bin/hbase org.apache.hadoop.hbase.io.hfile.HFile -v -f hdfs://10.81.47.41:8020/hbase/TEST/1418428042/DSMP/4759508618286845475 </code> </programlisting>If
you leave off the option -v to see just a summary on the hfile. See
usage for other things to do with the <classname>HFile</classname>
tool.</para>
</section>
<section xml:id="store.file.dir">
<title>StoreFile Directory Structure on HDFS</title>
<para>For more information of what StoreFiles look like on HDFS with respect to the directory structure, see <xref linkend="trouble.namenode.hbase.objects" />.
</para>
</section>
</section> <!-- hfile -->
<section xml:id="hfile.blocks">
<title>Blocks</title>
<para>StoreFiles are composed of blocks. The blocksize is configured on a per-ColumnFamily basis.
</para>
<para>Compression happens at the block level within StoreFiles. For more information on compression, see <xref linkend="compression"/>.
</para>
<para>For more information on blocks, see the <link xlink:href="http://hbase.apache.org/xref/org/apache/hadoop/hbase/io/hfile/HFileBlock.html">HFileBlock source code</link>.
</para>
</section>
<section xml:id="keyvalue">
<title>KeyValue</title>
<para>The KeyValue class is the heart of data storage in HBase. KeyValue wraps a byte array and takes offsets and lengths into passed array
at where to start interpreting the content as KeyValue.
</para>
<para>The KeyValue format inside a byte array is:
<itemizedlist>
<listitem>keylength</listitem>
<listitem>valuelength</listitem>
<listitem>key</listitem>
<listitem>value</listitem>
</itemizedlist>
</para>
<para>The Key is further decomposed as:
<itemizedlist>
<listitem>rowlength</listitem>
<listitem>row (i.e., the rowkey)</listitem>
<listitem>columnfamilylength</listitem>
<listitem>columnfamily</listitem>
<listitem>columnqualifier</listitem>
<listitem>timestamp</listitem>
<listitem>keytype (e.g., Put, Delete, DeleteColumn, DeleteFamily)</listitem>
</itemizedlist>
</para>
<para>KeyValue instances are <emphasis>not</emphasis> split across blocks.
For example, if there is an 8 MB KeyValue, even if the block-size is 64kb this KeyValue will be read
in as a coherent block. For more information, see the <link xlink:href="http://hbase.apache.org/xref/org/apache/hadoop/hbase/KeyValue.html">KeyValue source code</link>.
</para>
<section xml:id="keyvalue.example"><title>Example</title>
<para>To emphasize the points above, examine what happens with two Puts for two different columns for the same row:</para>
<itemizedlist>
<listitem>Put #1: <code>rowkey=row1, cf:attr1=value1</code></listitem>
<listitem>Put #2: <code>rowkey=row1, cf:attr2=value2</code></listitem>
</itemizedlist>
<para>Even though these are for the same row, a KeyValue is created for each column:</para>
<para>Key portion for Put #1:
<itemizedlist>
<listitem>rowlength <code>------------&gt; 4</code></listitem>
<listitem>row <code>-----------------&gt; row1</code></listitem>
<listitem>columnfamilylength <code>---&gt; 2</code></listitem>
<listitem>columnfamily <code>--------&gt; cf</code></listitem>
<listitem>columnqualifier <code>------&gt; attr1</code></listitem>
<listitem>timestamp <code>-----------&gt; server time of Put</code></listitem>
<listitem>keytype <code>-------------&gt; Put</code></listitem>
</itemizedlist>
</para>
<para>Key portion for Put #2:
<itemizedlist>
<listitem>rowlength <code>------------&gt; 4</code></listitem>
<listitem>row <code>-----------------&gt; row1</code></listitem>
<listitem>columnfamilylength <code>---&gt; 2</code></listitem>
<listitem>columnfamily <code>--------&gt; cf</code></listitem>
<listitem>columnqualifier <code>------&gt; attr2</code></listitem>
<listitem>timestamp <code>-----------&gt; server time of Put</code></listitem>
<listitem>keytype <code>-------------&gt; Put</code></listitem>
</itemizedlist>
<!-- trying to get the above arrows to line up at the same horizontal position has proven
harder than I thought -->
</para>
</section>
<para>It is critical to understand that the rowkey, ColumnFamily, and column (aka columnqualifier) are embedded within
the KeyValue instance. The longer these identifiers are, the bigger the KeyValue is.</para>
</section>
<section xml:id="compaction">
<title>Compaction</title>
<para>There are two types of compactions: minor and major. Minor compactions will usually pick up a couple of the smaller adjacent
StoreFiles and rewrite them as one. Minors do not drop deletes or expired cells, only major compactions do this. Sometimes a minor compaction
will pick up all the StoreFiles in the Store and in this case it actually promotes itself to being a major compaction.
</para>
<para>After a major compaction runs there will be a single StoreFile per Store, and this will help performance usually. Caution: major compactions rewrite all of the Stores data and on a loaded system, this may not be tenable;
major compactions will usually have to be done manually on large systems. See <xref linkend="managed.compactions" />.
</para>
<para>Compactions will <emphasis>not</emphasis> perform region merges. See <xref linkend="ops.regionmgt.merge"/> for more information on region merging.
</para>
<section xml:id="compaction.file.selection">
<title>Compaction File Selection</title>
<para>To understand the core algorithm for StoreFile selection, there is some ASCII-art in the <link xlink:href="http://hbase.apache.org/xref/org/apache/hadoop/hbase/regionserver/Store.html#836">Store source code</link> that
will serve as useful reference. It has been copied below:
<programlisting>
/* normal skew:
*
* older ----> newer
* _
* | | _
* | | | | _
* --|-|- |-|- |-|---_-------_------- minCompactSize
* | | | | | | | | _ | |
* | | | | | | | | | | | |
* | | | | | | | | | | | |
*/
</programlisting>
Important knobs:
<itemizedlist>
<listitem><code>hbase.store.compaction.ratio</code> Ratio used in compaction
file selection algorithm (default 1.2f). </listitem>
<listitem><code>hbase.hstore.compaction.min</code> (.90 hbase.hstore.compactionThreshold) (files) Minimum number
of StoreFiles per Store to be selected for a compaction to occur (default 2).</listitem>
<listitem><code>hbase.hstore.compaction.max</code> (files) Maximum number of StoreFiles to compact per minor compaction (default 10).</listitem>
<listitem><code>hbase.hstore.compaction.min.size</code> (bytes)
Any StoreFile smaller than this setting with automatically be a candidate for compaction. Defaults to
<code>hbase.hregion.memstore.flush.size</code> (128 mb). </listitem>
<listitem><code>hbase.hstore.compaction.max.size</code> (.92) (bytes)
Any StoreFile larger than this setting with automatically be excluded from compaction (default Long.MAX_VALUE). </listitem>
</itemizedlist>
</para>
<para>The minor compaction StoreFile selection logic is size based, and selects a file for compaction when the file
&lt;= sum(smaller_files) * <code>hbase.hstore.compaction.ratio</code>.
</para>
</section>
<section xml:id="compaction.file.selection.example1">
<title>Minor Compaction File Selection - Example #1 (Basic Example)</title>
<para>This example mirrors an example from the unit test <code>TestCompactSelection</code>.
<itemizedlist>
<listitem><code>hbase.store.compaction.ratio</code> = 1.0f </listitem>
<listitem><code>hbase.hstore.compaction.min</code> = 3 (files) </listitem>>
<listitem><code>hbase.hstore.compaction.max</code> = 5 (files) </listitem>>
<listitem><code>hbase.hstore.compaction.min.size</code> = 10 (bytes) </listitem>>
<listitem><code>hbase.hstore.compaction.max.size</code> = 1000 (bytes) </listitem>>
</itemizedlist>
The following StoreFiles exist: 100, 50, 23, 12, and 12 bytes apiece (oldest to newest).
With the above parameters, the files that would be selected for minor compaction are 23, 12, and 12.
</para>
<para>Why?
<itemizedlist>
<listitem>100 --&gt; No, because sum(50, 23, 12, 12) * 1.0 = 97. </listitem>
<listitem>50 --&gt; No, because sum(23, 12, 12) * 1.0 = 47. </listitem>
<listitem>23 --&gt; Yes, because sum(12, 12) * 1.0 = 24. </listitem>
<listitem>12 --&gt; Yes, because the previous file has been included, and because this
does not exceed the the max-file limit of 5 </listitem>
<listitem>12 --&gt; Yes, because the previous file had been included, and because this
does not exceed the the max-file limit of 5.</listitem>
</itemizedlist>
</para>
</section>
<section xml:id="compaction.file.selection.example2">
<title>Minor Compaction File Selection - Example #2 (Not Enough Files To Compact)</title>
<para>This example mirrors an example from the unit test <code>TestCompactSelection</code>.
<itemizedlist>
<listitem><code>hbase.store.compaction.ratio</code> = 1.0f </listitem>
<listitem><code>hbase.hstore.compaction.min</code> = 3 (files) </listitem>>
<listitem><code>hbase.hstore.compaction.max</code> = 5 (files) </listitem>>
<listitem><code>hbase.hstore.compaction.min.size</code> = 10 (bytes) </listitem>>
<listitem><code>hbase.hstore.compaction.max.size</code> = 1000 (bytes) </listitem>>
</itemizedlist>
</para>
<para>The following StoreFiles exist: 100, 25, 12, and 12 bytes apiece (oldest to newest).
With the above parameters, no compaction will be started.
</para>
<para>Why?
<itemizedlist>
<listitem>100 --&gt; No, because sum(25, 12, 12) * 1.0 = 47</listitem>
<listitem>25 --&gt; No, because sum(12, 12) * 1.0 = 24</listitem>
<listitem>12 --&gt; No. Candidate because sum(12) * 1.0 = 12, there are only 2 files to compact and that is less than the threshold of 3</listitem>
<listitem>12 --&gt; No. Candidate because the previous StoreFile was, but there are not enough files to compact</listitem>
</itemizedlist>
</para>
</section>
<section xml:id="compaction.file.selection.example2">
<title>Minor Compaction File Selection - Example #3 (Limiting Files To Compact)</title>
<para>This example mirrors an example from the unit test <code>TestCompactSelection</code>.
<itemizedlist>
<listitem><code>hbase.store.compaction.ratio</code> = 1.0f </listitem>
<listitem><code>hbase.hstore.compaction.min</code> = 3 (files) </listitem>>
<listitem><code>hbase.hstore.compaction.max</code> = 5 (files) </listitem>>
<listitem><code>hbase.hstore.compaction.min.size</code> = 10 (bytes) </listitem>>
<listitem><code>hbase.hstore.compaction.max.size</code> = 1000 (bytes) </listitem>>
</itemizedlist>
The following StoreFiles exist: 7, 6, 5, 4, 3, 2, and 1 bytes apiece (oldest to newest).
With the above parameters, the files that would be selected for minor compaction are 7, 6, 5, 4, 3.
</para>
<para>Why?
<itemizedlist>
<listitem>7 --&gt; Yes, because sum(6, 5, 4, 3, 2, 1) * 1.0 = 21. Also, 7 is less than the min-size</listitem>
<listitem>6 --&gt; Yes, because sum(5, 4, 3, 2, 1) * 1.0 = 15. Also, 6 is less than the min-size. </listitem>
<listitem>5 --&gt; Yes, because sum(4, 3, 2, 1) * 1.0 = 10. Also, 5 is less than the min-size. </listitem>
<listitem>4 --&gt; Yes, because sum(3, 2, 1) * 1.0 = 6. Also, 4 is less than the min-size. </listitem>
<listitem>3 --&gt; Yes, because sum(2, 1) * 1.0 = 3. Also, 3 is less than the min-size. </listitem>
<listitem>2 --&gt; No. Candidate because previous file was selected and 2 is less than the min-size, but the max-number of files to compact has been reached. </listitem>
<listitem>1 --&gt; No. Candidate because previous file was selected and 1 is less than the min-size, but max-number of files to compact has been reached. </listitem>
</itemizedlist>
</para>
</section>
<section xml:id="compaction.config.impact">
<title>Impact of Key Configuration Options</title>
<para><code>hbase.store.compaction.ratio</code>. A large ratio (e.g., 10) will produce a single giant file. Conversely, a value of .25 will
produce behavior similar to the BigTable compaction algorithm - resulting in 4 StoreFiles.
</para>
<para><code>hbase.hstore.compaction.min.size</code>. Because
this limit represents the "automatic include" limit for all StoreFiles smaller than this value, this value may need to
be adjusted downwards in write-heavy environments where many 1 or 2 mb StoreFiles are being flushed, because every file
will be targeted for compaction and the resulting files may still be under the min-size and require further compaction, etc.
</para>
</section>
<section xml:id="ops.stripe"><title>Experimental: stripe compactions</title>
<para>
Stripe compactions is an experimental feature added in HBase 0.98 which aims to improve compactions for large regions or non-uniformly distributed row keys. In order to achieve smaller and/or more granular compactions, the store files within a region are maintained separately for several row-key sub-ranges, or "stripes", of the region. The division is not visible to the higher levels of the system, so externally each region functions as before.
</para><para>
This feature is fully compatible with default compactions - it can be enabled for existing tables, and the table will continue to operate normally if it's disabled later.
</para>
</section><section xml:id="ops.stripe.when"><title>When to use</title>
<para>You might want to consider using this feature if you have:
<itemizedlist>
<listitem>
large regions (in that case, you can get the positive effect of much smaller regions without additional memstore and region management overhead); or
</listitem><listitem>
non-uniform row keys, e.g. time dimension in a key (in that case, only the stripes receiving the new keys will keep compacting - old data will not compact as much, or at all).
</listitem>
</itemizedlist>
</para><para>
According to perf testing performed, in these case the read performance can improve somewhat, and the read and write performance variability due to compactions is greatly reduced. There's overall perf improvement on large, non-uniform row key regions (hash-prefixed timestamp key) over long term. All of these performance gains are best realized when table is already large. In future, the perf improvement might also extend to region splits.
</para>
<section xml:id="ops.stripe.enable"><title>How to enable</title>
<para>
To use stripe compactions for a table or a column family, you should set its <varname>hbase.hstore.engine.class</varname> to <varname>org.apache.hadoop.hbase.regionserver.StripeStoreEngine</varname>. Due to the nature of compactions, you also need to set the blocking file count to a high number (100 is a good default, which is 10 times the normal default of 10). If changing the existing table, you should do it when it is disabled. Examples:
<programlisting>
alter 'orders_table', CONFIGURATION => {'hbase.hstore.engine.class' => 'org.apache.hadoop.hbase.regionserver.StripeStoreEngine', 'hbase.hstore.blockingStoreFiles' => '100'}
alter 'orders_table', {NAME => 'blobs_cf', CONFIGURATION => {'hbase.hstore.engine.class' => 'org.apache.hadoop.hbase.regionserver.StripeStoreEngine', 'hbase.hstore.blockingStoreFiles' => '100'}}
create 'orders_table', 'blobs_cf', CONFIGURATION => {'hbase.hstore.engine.class' => 'org.apache.hadoop.hbase.regionserver.StripeStoreEngine', 'hbase.hstore.blockingStoreFiles' => '100'}
</programlisting>
</para><para>
Then, you can configure the other options if needed (see below) and enable the table.
To switch back to default compactions, set <varname>hbase.hstore.engine.class</varname> to nil to unset it; or set it explicitly to "<varname>org.apache.hadoop.hbase.regionserver.DefaultStoreEngine</varname>" (this also needs to be done on a disabled table).
</para><para>
When you enable a large table after changing the store engine either way, a major compaction will likely be performed on most regions. This is not a problem with new tables.
</para>
</section><section xml:id="ops.stripe.config"><title>How to configure</title>
<para>
All of the settings described below are best set on table/cf level (with the table disabled first, for the settings to apply), similar to the above, e.g.
<programlisting>
alter 'orders_table', CONFIGURATION => {'key' => 'value', ..., 'key' => 'value'}}
</programlisting>
</para>
<section xml:id="ops.stripe.config.sizing"><title>Region and stripe sizing</title>
<para>
Based on your region sizing, you might want to also change your stripe sizing. By default, your new regions will start with one stripe. When the stripe is too big (16 memstore flushes size), on next compaction it will be split into two stripes. Stripe splitting will continue in a similar manner as the region grows, until the region itself is big enough to split (region split will work the same as with default compactions).
</para><para>
You can improve this pattern for your data. You should generally aim at stripe size of at least 1Gb, and about 8-12 stripes for uniform row keys - so, for example if your regions are 30 Gb, 12x2.5Gb stripes might be a good idea.
</para><para>
The settings are as follows:
<table frame='all'><tgroup cols='2' align='left' colsep='1' rowsep='1'><colspec colname='c1'/><colspec colname='c2'/>
<thead><row><entry>Setting</entry><entry>Notes</entry></row></thead>
<tbody>
<row><entry>
<varname>hbase.store.stripe.initialStripeCount</varname>
</entry><entry>
Initial stripe count to create. You can use it as follows:
<itemizedlist>
<listitem>
for relatively uniform row keys, if you know the approximate target number of stripes from the above, you can avoid some splitting overhead by starting w/several stripes (2, 5, 10...). Note that if the early data is not representative of overall row key distribution, this will not be as efficient.
</listitem><listitem>
for existing tables with lots of data, you can use this to pre-split stripes.
</listitem><listitem>
for e.g. hash-prefixed sequential keys, with more than one hash prefix per region, you know that some pre-splitting makes sense.
</listitem>
</itemizedlist>
</entry></row><row><entry>
<varname>hbase.store.stripe.sizeToSplit</varname>
</entry><entry>
Maximum stripe size before it's split. You can use this in conjunction with the next setting to control target stripe size (sizeToSplit = splitPartsCount * target stripe size), according to the above sizing considerations.
</entry></row><row><entry>
<varname>hbase.store.stripe.splitPartCount</varname>
</entry><entry>
The number of new stripes to create when splitting one. The default is 2, and is good for most cases. For non-uniform row keys, you might experiment with increasing the number somewhat (3-4), to isolate the arriving updates into narrower slice of the region with just one split instead of several.
</entry></row>
</tbody>
</tgroup></table>
</para>
</section><section xml:id="ops.stripe.config.memstore"><title>Memstore sizing</title>
<para>
By default, the flush creates several files from one memstore, according to existing stripe boundaries and row keys to flush. This approach minimizes write amplification, but can be undesirable if memstore is small and there are many stripes (the files will be too small).
</para><para>
In such cases, you can set <varname>hbase.store.stripe.compaction.flushToL0</varname> to true. This will cause flush to create a single file instead; when at least <varname>hbase.store.stripe.compaction.minFilesL0</varname> such files (by default, 4) accumulate, they will be compacted into striped files.</para>
</section><section xml:id="ops.stripe.config.compact"><title>Normal compaction configuration</title>
<para>
All the settings that apply to normal compactions (file size limits, etc.) apply to stripe compactions. The exception are min and max number of files, which are set to higher values by default because the files in stripes are smaller. To control these for stripe compactions, use <varname>hbase.store.stripe.compaction.minFiles</varname> and <varname>.maxFiles</varname>.
</para>
</section>
</section>
</section>
</section> <!-- compaction -->
</section> <!-- store -->
</section> <!-- regions -->
<section xml:id="arch.bulk.load"><title>Bulk Loading</title>
<section xml:id="arch.bulk.load.overview"><title>Overview</title>
<para>
HBase includes several methods of loading data into tables.
The most straightforward method is to either use the <code>TableOutputFormat</code>
class from a MapReduce job, or use the normal client APIs; however,
these are not always the most efficient methods.
</para>
<para>
The bulk load feature uses a MapReduce job to output table data in HBase's internal
data format, and then directly loads the generated StoreFiles into a running
cluster. Using bulk load will use less CPU and network resources than
simply using the HBase API.
</para>
</section>
<section xml:id="arch.bulk.load.limitations"><title>Bulk Load Limitations</title>
<para>As bulk loading bypasses the write path, the WAL doesn’t get written to as part of the process.
Replication works by reading the WAL files so it won’t see the bulk loaded data – and the same goes for the edits that use Put.setWriteToWAL(true).
One way to handle that is to ship the raw files or the HFiles to the other cluster and do the other processing there.</para>
</section>
<section xml:id="arch.bulk.load.arch"><title>Bulk Load Architecture</title>
<para>
The HBase bulk load process consists of two main steps.
</para>
<section xml:id="arch.bulk.load.prep"><title>Preparing data via a MapReduce job</title>
<para>
The first step of a bulk load is to generate HBase data files (StoreFiles) from
a MapReduce job using <code>HFileOutputFormat</code>. This output format writes
out data in HBase's internal storage format so that they can be
later loaded very efficiently into the cluster.
</para>
<para>
In order to function efficiently, <code>HFileOutputFormat</code> must be
configured such that each output HFile fits within a single region.
In order to do this, jobs whose output will be bulk loaded into HBase
use Hadoop's <code>TotalOrderPartitioner</code> class to partition the map output
into disjoint ranges of the key space, corresponding to the key
ranges of the regions in the table.
</para>
<para>
<code>HFileOutputFormat</code> includes a convenience function,
<code>configureIncrementalLoad()</code>, which automatically sets up
a <code>TotalOrderPartitioner</code> based on the current region boundaries of a
table.
</para>
</section>
<section xml:id="arch.bulk.load.complete"><title>Completing the data load</title>
<para>
After the data has been prepared using
<code>HFileOutputFormat</code>, it is loaded into the cluster using
<code>completebulkload</code>. This command line tool iterates
through the prepared data files, and for each one determines the
region the file belongs to. It then contacts the appropriate Region
Server which adopts the HFile, moving it into its storage directory
and making the data available to clients.
</para>
<para>
If the region boundaries have changed during the course of bulk load
preparation, or between the preparation and completion steps, the
<code>completebulkloads</code> utility will automatically split the
data files into pieces corresponding to the new boundaries. This
process is not optimally efficient, so users should take care to
minimize the delay between preparing a bulk load and importing it
into the cluster, especially if other clients are simultaneously
loading data through other means.
</para>
</section>
</section>
<section xml:id="arch.bulk.load.import"><title>Importing the prepared data using the completebulkload tool</title>
<para>
After a data import has been prepared, either by using the
<code>importtsv</code> tool with the
"<code>importtsv.bulk.output</code>" option or by some other MapReduce
job using the <code>HFileOutputFormat</code>, the
<code>completebulkload</code> tool is used to import the data into the
running cluster.
</para>
<para>
The <code>completebulkload</code> tool simply takes the output path
where <code>importtsv</code> or your MapReduce job put its results, and
the table name to import into. For example:
</para>
<code>$ hadoop jar hbase-VERSION.jar completebulkload [-c /path/to/hbase/config/hbase-site.xml] /user/todd/myoutput mytable</code>
<para>
The <code>-c config-file</code> option can be used to specify a file
containing the appropriate hbase parameters (e.g., hbase-site.xml) if
not supplied already on the CLASSPATH (In addition, the CLASSPATH must
contain the directory that has the zookeeper configuration file if
zookeeper is NOT managed by HBase).
</para>
<para>
Note: If the target table does not already exist in HBase, this
tool will create the table automatically.</para>
<para>
This tool will run quickly, after which point the new data will be visible in
the cluster.
</para>
</section>
<section xml:id="arch.bulk.load.also"><title>See Also</title>
<para>For more information about the referenced utilities, see <xref linkend="importtsv"/> and <xref linkend="completebulkload"/>.
</para>
<para>
See <link xlink:ref="http://blog.cloudera.com/blog/2013/09/how-to-use-hbase-bulk-loading-and-why/">How-to: Use HBase Bulk Loading, and Why</link>
for a recent blog on current state of bulk loading.
</para>
</section>
<section xml:id="arch.bulk.load.adv"><title>Advanced Usage</title>
<para>
Although the <code>importtsv</code> tool is useful in many cases, advanced users may
want to generate data programatically, or import data from other formats. To get
started doing so, dig into <code>ImportTsv.java</code> and check the JavaDoc for
HFileOutputFormat.
</para>
<para>
The import step of the bulk load can also be done programatically. See the
<code>LoadIncrementalHFiles</code> class for more information.
</para>
</section>
</section> <!-- bulk loading -->
<section xml:id="arch.hdfs"><title>HDFS</title>
<para>As HBase runs on HDFS (and each StoreFile is written as a file on HDFS),
it is important to have an understanding of the HDFS Architecture
especially in terms of how it stores files, handles failovers, and replicates blocks.
</para>
<para>See the Hadoop documentation on <link xlink:href="http://hadoop.apache.org/common/docs/current/hdfs_design.html">HDFS Architecture</link>
for more information.
</para>
<section xml:id="arch.hdfs.nn"><title>NameNode</title>
<para>The NameNode is responsible for maintaining the filesystem metadata. See the above HDFS Architecture link
for more information.
</para>
</section>
<section xml:id="arch.hdfs.dn"><title>DataNode</title>
<para>The DataNodes are responsible for storing HDFS blocks. See the above HDFS Architecture link
for more information.
</para>
</section>
</section>
</chapter> <!-- architecture -->
<xi:include xmlns:xi="http://www.w3.org/2001/XInclude" href="external_apis.xml" />
<xi:include xmlns:xi="http://www.w3.org/2001/XInclude" href="cp.xml" />
<xi:include xmlns:xi="http://www.w3.org/2001/XInclude" href="performance.xml" />
<xi:include xmlns:xi="http://www.w3.org/2001/XInclude" href="troubleshooting.xml" />
<xi:include xmlns:xi="http://www.w3.org/2001/XInclude" href="case_studies.xml" />
<xi:include xmlns:xi="http://www.w3.org/2001/XInclude" href="ops_mgt.xml" />
<xi:include xmlns:xi="http://www.w3.org/2001/XInclude" href="developer.xml" />
<xi:include xmlns:xi="http://www.w3.org/2001/XInclude" href="zookeeper.xml" />
<xi:include xmlns:xi="http://www.w3.org/2001/XInclude" href="cp.xml" />
<xi:include xmlns:xi="http://www.w3.org/2001/XInclude" href="community.xml" />
<appendix xml:id="faq">
<title >FAQ</title>
<qandaset defaultlabel='faq'>
<qandadiv><title>General</title>
<qandaentry>
<question><para>When should I use HBase?</para></question>
<answer>
<para>See the <xref linkend="arch.overview" /> in the Architecture chapter.
</para>
</answer>
</qandaentry>
<qandaentry>
<question><para>Are there other HBase FAQs?</para></question>
<answer>
<para>
See the FAQ that is up on the wiki, <link xlink:href="http://wiki.apache.org/hadoop/Hbase/FAQ">HBase Wiki FAQ</link>.
</para>
</answer>
</qandaentry>
<qandaentry xml:id="faq.sql">
<question><para>Does HBase support SQL?</para></question>
<answer>
<para>
Not really. SQL-ish support for HBase via <link xlink:href="http://hive.apache.org/">Hive</link> is in development, however Hive is based on MapReduce which is not generally suitable for low-latency requests.
See the <xref linkend="datamodel" /> section for examples on the HBase client.
</para>
</answer>
</qandaentry>
<qandaentry>
<question><para>How can I find examples of NoSQL/HBase?</para></question>
<answer>
<para>See the link to the BigTable paper in <xref linkend="other.info" /> in the appendix, as
well as the other papers.
</para>
</answer>
</qandaentry>
<qandaentry>
<question><para>What is the history of HBase?</para></question>
<answer>
<para>See <xref linkend="hbase.history"/>.
</para>
</answer>
</qandaentry>
</qandadiv>
<qandadiv xml:id="faq.arch"><title>Architecture</title>
<qandaentry xml:id="faq.arch.regions">
<question><para>How does HBase handle Region-RegionServer assignment and locality?</para></question>
<answer>
<para>
See <xref linkend="regions.arch" />.
</para>
</answer>
</qandaentry>
</qandadiv>
<qandadiv xml:id="faq.config"><title>Configuration</title>
<qandaentry xml:id="faq.config.started">
<question><para>How can I get started with my first cluster?</para></question>
<answer>
<para>
See <xref linkend="quickstart" />.
</para>
</answer>
</qandaentry>
<qandaentry xml:id="faq.config.started">
<question><para>Where can I learn about the rest of the configuration options?</para></question>
<answer>
<para>
See <xref linkend="configuration" />.
</para>
</answer>
</qandaentry>
</qandadiv>
<qandadiv xml:id="faq.design"><title>Schema Design / Data Access</title>
<qandaentry xml:id="faq.design.schema">
<question><para>How should I design my schema in HBase?</para></question>
<answer>
<para>
See <xref linkend="datamodel" /> and <xref linkend="schema" />
</para>
</answer>
</qandaentry>
<qandaentry>
<question><para>
How can I store (fill in the blank) in HBase?
</para></question>
<answer>
<para>
See <xref linkend="supported.datatypes" />.
</para>
</answer>
</qandaentry>
<qandaentry xml:id="secondary.indices">
<question><para>
How can I handle secondary indexes in HBase?
</para></question>
<answer>
<para>
See <xref linkend="secondary.indexes" />
</para>
</answer>
</qandaentry>
<qandaentry xml:id="faq.changing.rowkeys">
<question><para>Can I change a table's rowkeys?</para></question>
<answer>
<para>
This is a very common quesiton. You can't. See <xref linkend="changing.rowkeys" />.
</para>
</answer>
</qandaentry>
<qandaentry xml:id="faq.apis">
<question><para>What APIs does HBase support?</para></question>
<answer>
<para>
See <xref linkend="datamodel" />, <xref linkend="client" /> and <xref linkend="nonjava.jvm"/>.
</para>
</answer>
</qandaentry>
</qandadiv>
<qandadiv xml:id="faq.mapreduce"><title>MapReduce</title>
<qandaentry xml:id="faq.mapreduce.use">
<question><para>How can I use MapReduce with HBase?</para></question>
<answer>
<para>
See <xref linkend="mapreduce" />
</para>
</answer>
</qandaentry>
</qandadiv>
<qandadiv><title>Performance and Troubleshooting</title>
<qandaentry>
<question><para>
How can I improve HBase cluster performance?
</para></question>
<answer>
<para>
See <xref linkend="performance" />.
</para>
</answer>
</qandaentry>
<qandaentry>
<question><para>
How can I troubleshoot my HBase cluster?
</para></question>
<answer>
<para>
See <xref linkend="trouble" />.
</para>
</answer>
</qandaentry>
</qandadiv>
<qandadiv xml:id="ec2"><title>Amazon EC2</title>
<qandaentry>
<question><para>
I am running HBase on Amazon EC2 and...
</para></question>
<answer>
<para>
EC2 issues are a special case. See Troubleshooting <xref linkend="trouble.ec2" /> and Performance <xref linkend="perf.ec2" /> sections.
</para>
</answer>
</qandaentry>
</qandadiv>
<qandadiv><title xml:id="faq.operations">Operations</title>
<qandaentry>
<question><para>
How do I manage my HBase cluster?
</para></question>
<answer>
<para>
See <xref linkend="ops_mgt" />
</para>
</answer>
</qandaentry>
<qandaentry>
<question><para>
How do I back up my HBase cluster?
</para></question>
<answer>
<para>
See <xref linkend="ops.backup" />
</para>
</answer>
</qandaentry>
</qandadiv>
<qandadiv><title>HBase in Action</title>
<qandaentry>
<question><para>Where can I find interesting videos and presentations on HBase?</para></question>
<answer>
<para>
See <xref linkend="other.info" />
</para>
</answer>
</qandaentry>
</qandadiv>
</qandaset>
</appendix>
<appendix xml:id="hbck.in.depth">
<title>hbck In Depth</title>
<para>HBaseFsck (hbck) is a tool for checking for region consistency and table integrity problems
and repairing a corrupted HBase. It works in two basic modes -- a read-only inconsistency
identifying mode and a multi-phase read-write repair mode.
</para>
<section>
<title>Running hbck to identify inconsistencies</title>
To check to see if your HBase cluster has corruptions, run hbck against your HBase cluster:
<programlisting>
$ ./bin/hbase hbck
</programlisting>
<para>
At the end of the commands output it prints OK or tells you the number of INCONSISTENCIES
present. You may also want to run run hbck a few times because some inconsistencies can be
transient (e.g. cluster is starting up or a region is splitting). Operationally you may want to run
hbck regularly and setup alert (e.g. via nagios) if it repeatedly reports inconsistencies .
A run of hbck will report a list of inconsistencies along with a brief description of the regions and
tables affected. The using the <code>-details</code> option will report more details including a representative
listing of all the splits present in all the tables.
</para>
<programlisting>
$ ./bin/hbase hbck -details
</programlisting>
If you just want to know if some tables are corrupted, you can limit hbck to identify inconsistencies
in only specific tables. For example the following command would only attempt to check table
TableFoo and TableBar. The benefit is that hbck will run in less time.
<programlisting>
$ ./bin/hbase hbck TableFoo TableBar
</programlisting>
</section>
<section><title>Inconsistencies</title>
<para>
If after several runs, inconsistencies continue to be reported, you may have encountered a
corruption. These should be rare, but in the event they occur newer versions of HBase include
the hbck tool enabled with automatic repair options.
</para>
<para>
There are two invariants that when violated create inconsistencies in HBase:
</para>
<itemizedlist>
<listitem>HBase’s region consistency invariant is satisfied if every region is assigned and
deployed on exactly one region server, and all places where this state kept is in
accordance.
</listitem>
<listitem>HBase’s table integrity invariant is satisfied if for each table, every possible row key
resolves to exactly one region.
</listitem>
</itemizedlist>
<para>
Repairs generally work in three phases -- a read-only information gathering phase that identifies
inconsistencies, a table integrity repair phase that restores the table integrity invariant, and then
finally a region consistency repair phase that restores the region consistency invariant.
Starting from version 0.90.0, hbck could detect region consistency problems report on a subset
of possible table integrity problems. It also included the ability to automatically fix the most
common inconsistency, region assignment and deployment consistency problems. This repair
could be done by using the <code>-fix</code> command line option. These problems close regions if they are
open on the wrong server or on multiple region servers and also assigns regions to region
servers if they are not open.
</para>
<para>
Starting from HBase versions 0.90.7, 0.92.2 and 0.94.0, several new command line options are
introduced to aid repairing a corrupted HBase. This hbck sometimes goes by the nickname
“uberhbck”. Each particular version of uber hbck is compatible with the HBase’s of the same
major version (0.90.7 uberhbck can repair a 0.90.4). However, versions &lt;=0.90.6 and versions
&lt;=0.92.1 may require restarting the master or failing over to a backup master.
</para>
</section>
<section><title>Localized repairs</title>
<para>
When repairing a corrupted HBase, it is best to repair the lowest risk inconsistencies first.
These are generally region consistency repairs -- localized single region repairs, that only modify
in-memory data, ephemeral zookeeper data, or patch holes in the META table.
Region consistency requires that the HBase instance has the state of the region’s data in HDFS
(.regioninfo files), the region’s row in the .META. table., and region’s deployment/assignments on
region servers and the master in accordance. Options for repairing region consistency include:
<itemizedlist>
<listitem><code>-fixAssignments</code> (equivalent to the 0.90 <code>-fix</code> option) repairs unassigned, incorrectly
assigned or multiply assigned regions.
</listitem>
<listitem><code>-fixMeta</code> which removes meta rows when corresponding regions are not present in
HDFS and adds new meta rows if they regions are present in HDFS while not in META.
</listitem>
</itemizedlist>
To fix deployment and assignment problems you can run this command:
</para>
<programlisting>
$ ./bin/hbase hbck -fixAssignments
</programlisting>
To fix deployment and assignment problems as well as repairing incorrect meta rows you can
run this command:.
<programlisting>
$ ./bin/hbase hbck -fixAssignments -fixMeta
</programlisting>
There are a few classes of table integrity problems that are low risk repairs. The first two are
degenerate (startkey == endkey) regions and backwards regions (startkey > endkey). These are
automatically handled by sidelining the data to a temporary directory (/hbck/xxxx).
The third low-risk class is hdfs region holes. This can be repaired by using the:
<itemizedlist>
<listitem><code>-fixHdfsHoles</code> option for fabricating new empty regions on the file system.
If holes are detected you can use -fixHdfsHoles and should include -fixMeta and -fixAssignments to make the new region consistent.
</listitem>
</itemizedlist>
<programlisting>
$ ./bin/hbase hbck -fixAssignments -fixMeta -fixHdfsHoles
</programlisting>
Since this is a common operation, we’ve added a the <code>-repairHoles</code> flag that is equivalent to the
previous command:
<programlisting>
$ ./bin/hbase hbck -repairHoles
</programlisting>
If inconsistencies still remain after these steps, you most likely have table integrity problems
related to orphaned or overlapping regions.
</section>
<section><title>Region Overlap Repairs</title>
Table integrity problems can require repairs that deal with overlaps. This is a riskier operation
because it requires modifications to the file system, requires some decision making, and may
require some manual steps. For these repairs it is best to analyze the output of a <code>hbck -details</code>
run so that you isolate repairs attempts only upon problems the checks identify. Because this is
riskier, there are safeguard that should be used to limit the scope of the repairs.
WARNING: This is a relatively new and have only been tested on online but idle HBase instances
(no reads/writes). Use at your own risk in an active production environment!
The options for repairing table integrity violations include:
<itemizedlist>
<listitem><code>-fixHdfsOrphans</code> option for “adopting” a region directory that is missing a region
metadata file (the .regioninfo file).
</listitem>
<listitem><code>-fixHdfsOverlaps</code> ability for fixing overlapping regions
</listitem>
</itemizedlist>
When repairing overlapping regions, a region’s data can be modified on the file system in two
ways: 1) by merging regions into a larger region or 2) by sidelining regions by moving data to
“sideline” directory where data could be restored later. Merging a large number of regions is
technically correct but could result in an extremely large region that requires series of costly
compactions and splitting operations. In these cases, it is probably better to sideline the regions
that overlap with the most other regions (likely the largest ranges) so that merges can happen on
a more reasonable scale. Since these sidelined regions are already laid out in HBase’s native
directory and HFile format, they can be restored by using HBase’s bulk load mechanism.
The default safeguard thresholds are conservative. These options let you override the default
thresholds and to enable the large region sidelining feature.
<itemizedlist>
<listitem><code>-maxMerge &lt;n&gt;</code> maximum number of overlapping regions to merge
</listitem>
<listitem><code>-sidelineBigOverlaps</code> if more than maxMerge regions are overlapping, sideline attempt
to sideline the regions overlapping with the most other regions.
</listitem>
<listitem><code>-maxOverlapsToSideline &lt;n&gt;</code> if sidelining large overlapping regions, sideline at most n
regions.
</listitem>
</itemizedlist>
Since often times you would just want to get the tables repaired, you can use this option to turn
on all repair options:
<itemizedlist>
<listitem><code>-repair</code> includes all the region consistency options and only the hole repairing table
integrity options.
</listitem>
</itemizedlist>
Finally, there are safeguards to limit repairs to only specific tables. For example the following
command would only attempt to check and repair table TableFoo and TableBar.
<programlisting>
$ ./bin/hbase hbck -repair TableFoo TableBar
</programlisting>
<section><title>Special cases: Meta is not properly assigned</title>
There are a few special cases that hbck can handle as well.
Sometimes the meta table’s only region is inconsistently assigned or deployed. In this case
there is a special <code>-fixMetaOnly</code> option that can try to fix meta assignments.
<programlisting>
$ ./bin/hbase hbck -fixMetaOnly -fixAssignments
</programlisting>
</section>
<section><title>Special cases: HBase version file is missing</title>
HBase’s data on the file system requires a version file in order to start. If this flie is missing, you
can use the <code>-fixVersionFile</code> option to fabricating a new HBase version file. This assumes that
the version of hbck you are running is the appropriate version for the HBase cluster.
</section>
<section><title>Special case: Root and META are corrupt.</title>
The most drastic corruption scenario is the case where the ROOT or META is corrupted and
HBase will not start. In this case you can use the OfflineMetaRepair tool create new ROOT
and META regions and tables.
This tool assumes that HBase is offline. It then marches through the existing HBase home
directory, loads as much information from region metadata files (.regioninfo files) as possible
from the file system. If the region metadata has proper table integrity, it sidelines the original root
and meta table directories, and builds new ones with pointers to the region directories and their
data.
<programlisting>
$ ./bin/hbase org.apache.hadoop.hbase.util.hbck.OfflineMetaRepair
</programlisting>
NOTE: This tool is not as clever as uberhbck but can be used to bootstrap repairs that uberhbck
can complete.
If the tool succeeds you should be able to start hbase and run online repairs if necessary.
</section>
<section><title>Special cases: Offline split parent</title>
<para>
Once a region is split, the offline parent will be cleaned up automatically. Sometimes, daughter regions
are split again before their parents are cleaned up. HBase can clean up parents in the right order. However,
there could be some lingering offline split parents sometimes. They are in META, in HDFS, and not deployed.
But HBase can't clean them up. In this case, you can use the <code>-fixSplitParents</code> option to reset
them in META to be online and not split. Therefore, hbck can merge them with other regions if fixing
overlapping regions option is used.
</para>
<para>
This option should not normally be used, and it is not in <code>-fixAll</code>.
</para>
</section>
</section>
</appendix>
<appendix xml:id="compression">
<title >Compression In HBase<indexterm><primary>Compression</primary></indexterm></title>
<para>There are a bunch of compression options in HBase. There is some helpful discussion
to be found in <link xlink:href="http://search-hadoop.com/m/lL12B1PFVhp1">Documenting Guidance on compression and codecs</link>.
</para>
<section xml:id="compression.test">
<title>CompressionTest Tool</title>
<para>
HBase includes a tool to test compression is set up properly.
To run it, type <code>/bin/hbase org.apache.hadoop.hbase.util.CompressionTest</code>.
This will emit usage on how to run the tool.
</para>
<note><title>You need to restart regionserver for it to pick up fixed codecs!</title>
<para>Be aware that the regionserver caches the result of the compression check it runs
ahead of each region open. This means
that you will have to restart the regionserver for it to notice that you have fixed
any codec issues.</para>
</note>
</section>
<section xml:id="hbase.regionserver.codecs">
<title>
<varname>
hbase.regionserver.codecs
</varname>
</title>
<para>
To have a RegionServer test a set of codecs and fail-to-start if any
code is missing or misinstalled, add the configuration
<varname>
hbase.regionserver.codecs
</varname>
to your <filename>hbase-site.xml</filename> with a value of
codecs to test on startup. For example if the
<varname>
hbase.regionserver.codecs
</varname> value is <code>lzo,gz</code> and if lzo is not present
or improperly installed, the misconfigured RegionServer will fail
to start.
</para>
<para>
Administrators might make use of this facility to guard against
the case where a new server is added to cluster but the cluster
requires install of a particular coded.
</para>
</section>
<section xml:id="lzo.compression">
<title>
LZO
</title>
<para>Unfortunately, HBase cannot ship with LZO because of
the licensing issues; HBase is Apache-licensed, LZO is GPL.
Therefore LZO install is to be done post-HBase install.
See the <link xlink:href="http://wiki.apache.org/hadoop/UsingLzoCompression">Using LZO Compression</link>
wiki page for how to make LZO work with HBase.
</para>
<para>A common problem users run into when using LZO is that while initial
setup of the cluster runs smooth, a month goes by and some sysadmin goes to
add a machine to the cluster only they'll have forgotten to do the LZO
fixup on the new machine. In versions since HBase 0.90.0, we should
fail in a way that makes it plain what the problem is, but maybe not. </para>
<para>See <xref linkend="hbase.regionserver.codecs" />
for a feature to help protect against failed LZO install.</para>
</section>
<section xml:id="gzip.compression">
<title>
GZIP
</title>
<para>
GZIP will generally compress better than LZO though slower.
For some setups, better compression may be preferred.
Java will use java's GZIP unless the native Hadoop libs are
available on the CLASSPATH; in this case it will use native
compressors instead (If the native libs are NOT present,
you will see lots of <emphasis>Got brand-new compressor</emphasis>
reports in your logs; see <xref linkend="brand.new.compressor" />).
</para>
</section>
<section xml:id="snappy.compression">
<title>
SNAPPY
</title>
<para>
If snappy is installed, HBase can make use of it (courtesy of
<link xlink:href="http://code.google.com/p/hadoop-snappy/">hadoop-snappy</link>
<footnote><para>See <link xlink:href="http://search-hadoop.com/m/Ds8d51c263B1/%2522Hadoop-Snappy+in+synch+with+Hadoop+trunk%2522&amp;subj=Hadoop+Snappy+in+synch+with+Hadoop+trunk">Alejandro's note</link> up on the list on difference between Snappy in Hadoop
and Snappy in HBase</para></footnote>).
<orderedlist>
<listitem>
<para>
Build and install <link xlink:href="http://code.google.com/p/snappy/">snappy</link> on all nodes
of your cluster (see below). HBase nor Hadoop cannot include snappy because of licensing issues (The
hadoop libhadoop.so under its native dir does not include snappy; of note, the shipped .so
may be for 32-bit architectures -- this fact has tripped up folks in the past with them thinking
it 64-bit). The notes below are about installing snappy for HBase use. You may want snappy
available in your hadoop context also. That is not covered here.
HBase and Hadoop find the snappy .so in different locations currently: Hadoop picks those files in
<filename>./lib</filename> while HBase finds the .so in <filename>./lib/[PLATFORM]</filename>.
</para>
</listitem>
<listitem>
<para>
Use CompressionTest to verify snappy support is enabled and the libs can be loaded ON ALL NODES of your cluster:
<programlisting>$ hbase org.apache.hadoop.hbase.util.CompressionTest hdfs://host/path/to/hbase snappy</programlisting>
</para>
</listitem>
<listitem>
<para>
Create a column family with snappy compression and verify it in the hbase shell:
<programlisting>$ hbase> create 't1', { NAME => 'cf1', COMPRESSION => 'SNAPPY' }
hbase> describe 't1'</programlisting>
In the output of the "describe" command, you need to ensure it lists "COMPRESSION => 'SNAPPY'"
</para>
</listitem>
</orderedlist>
</para>
<section xml:id="snappy.compression.installation">
<title>
Installation
</title>
<para>Snappy is used by hbase to compress HFiles on flush and when compacting.
</para>
<para>
You will find the snappy library file under the .libs directory from your Snappy build (For example
/home/hbase/snappy-1.0.5/.libs/). The file is called libsnappy.so.1.x.x where 1.x.x is the version of the snappy
code you are building. You can either copy this file into your hbase lib directory -- under lib/native/PLATFORM --
naming the file as libsnappy.so,
or simply create a symbolic link to it (See ./bin/hbase for how it does library path for native libs).
</para>
<para>
The second file you need is the hadoop native library. You will find this file in your hadoop installation directory
under lib/native/Linux-amd64-64/ or lib/native/Linux-i386-32/. The file you are looking for is libhadoop.so.1.x.x.
Again, you can simply copy this file or link to it from under hbase in lib/native/PLATFORM (e.g. Linux-amd64-64, etc.),
using the name libhadoop.so.
</para>
<para>
At the end of the installation, you should have both libsnappy.so and libhadoop.so links or files present into
lib/native/Linux-amd64-64 or into lib/native/Linux-i386-32 (where the last part of the directory path is the
PLATFORM you built and rare running the native lib on)
</para>
<para>To point hbase at snappy support, in hbase-env.sh set
<programlisting>export HBASE_LIBRARY_PATH=/pathtoyourhadoop/lib/native/Linux-amd64-64</programlisting>
In <filename>/pathtoyourhadoop/lib/native/Linux-amd64-64</filename> you should have something like:
<programlisting>
libsnappy.a
libsnappy.so
libsnappy.so.1
libsnappy.so.1.1.2
</programlisting>
</para>
</section>
</section>
<section xml:id="changing.compression">
<title>Changing Compression Schemes</title>
<para>A frequent question on the dist-list is how to change compression schemes for ColumnFamilies. This is actually quite simple,
and can be done via an alter command. Because the compression scheme is encoded at the block-level in StoreFiles, the table does
<emphasis>not</emphasis> need to be re-created and the data does <emphasis>not</emphasis> copied somewhere else. Just make sure
the old codec is still available until you are sure that all of the old StoreFiles have been compacted.
</para>
</section>
</appendix>
<appendix>
<title xml:id="ycsb"><link xlink:href="https://github.com/brianfrankcooper/YCSB/">YCSB: The Yahoo! Cloud Serving Benchmark</link> and HBase</title>
<para>TODO: Describe how YCSB is poor for putting up a decent cluster load.</para>
<para>TODO: Describe setup of YCSB for HBase. In particular, presplit your tables before you start
a run. See <link xlink:href="https://issues.apache.org/jira/browse/HBASE-4163">HBASE-4163 Create Split Strategy for YCSB Benchmark</link>
for why and a little shell command for how to do it.</para>
<para>Ted Dunning redid YCSB so it's mavenized and added facility for verifying workloads. See <link xlink:href="https://github.com/tdunning/YCSB">Ted Dunning's YCSB</link>.</para>
</appendix>
<appendix xml:id="hfilev2">
<title>HFile format version 2</title>
<section><title>Motivation </title>
<para>Note: this feature was introduced in HBase 0.92</para>
<para>We found it necessary to revise the HFile format after encountering high memory usage and slow startup times caused by large Bloom filters and block indexes in the region server. Bloom filters can get as large as 100 MB per HFile, which adds up to 2 GB when aggregated over 20 regions. Block indexes can grow as large as 6 GB in aggregate size over the same set of regions. A region is not considered opened until all of its block index data is loaded. Large Bloom filters produce a different performance problem: the first get request that requires a Bloom filter lookup will incur the latency of loading the entire Bloom filter bit array.</para>
<para>To speed up region server startup we break Bloom filters and block indexes into multiple blocks and write those blocks out as they fill up, which also reduces the HFile writer’s memory footprint. In the Bloom filter case, “filling up a block” means accumulating enough keys to efficiently utilize a fixed-size bit array, and in the block index case we accumulate an “index block” of the desired size. Bloom filter blocks and index blocks (we call these “inline blocks”) become interspersed with data blocks, and as a side effect we can no longer rely on the difference between block offsets to determine data block length, as it was done in version 1.</para>
<para>HFile is a low-level file format by design, and it should not deal with application-specific details such as Bloom filters, which are handled at StoreFile level. Therefore, we call Bloom filter blocks in an HFile "inline" blocks. We also supply HFile with an interface to write those inline blocks. </para>
<para>Another format modification aimed at reducing the region server startup time is to use a contiguous “load-on-open” section that has to be loaded in memory at the time an HFile is being opened. Currently, as an HFile opens, there are separate seek operations to read the trailer, data/meta indexes, and file info. To read the Bloom filter, there are two more seek operations for its “data” and “meta” portions. In version 2, we seek once to read the trailer and seek again to read everything else we need to open the file from a contiguous block.</para></section>
<section><title>HFile format version 1 overview </title><para>As we will be discussing the changes we are making to the HFile format, it is useful to give a short overview of the previous (HFile version 1) format. An HFile in the existing format is structured as follows:
<inlinemediaobject>
<imageobject>
<imagedata align="middle" valign="middle" fileref="hfile.png" />
</imageobject>
<textobject>
<phrase>HFile Version 1</phrase>
</textobject>
<caption>
<para>HFile Version 1
</para>
</caption>
</inlinemediaobject>
<footnote><para>Image courtesy of Lars George, <link xlink:href="http://www.larsgeorge.com/2009/10/hbase-architecture-101-storage.html">hbase-architecture-101-storage.html</link>.</para></footnote>
</para>
<section><title> Block index format in version 1 </title>
<para>The block index in version 1 is very straightforward. For each entry, it contains: </para>
<orderedlist>
<listitem>
<para>Offset (long)</para>
</listitem>
<listitem>
<para>Uncompressed size (int)</para>
</listitem>
<listitem>
<para>Key (a serialized byte array written using Bytes.writeByteArray) </para>
<orderedlist>
<listitem>
<para>Key length as a variable-length integer (VInt)
</para>
</listitem>
<listitem>
<para>
Key bytes
</para>
</listitem>
</orderedlist>
</listitem>
</orderedlist>
<para>The number of entries in the block index is stored in the fixed file trailer, and has to be passed in to the method that reads the block index. One of the limitations of the block index in version 1 is that it does not provide the compressed size of a block, which turns out to be necessary for decompression. Therefore, the HFile reader has to infer this compressed size from the offset difference between blocks. We fix this limitation in version 2, where we store on-disk block size instead of uncompressed size, and get uncompressed size from the block header.</para></section></section><section><title>
HBase file format with inline blocks (version 2)
</title>
<section><title> Overview</title>
<para>The version of HBase introducing the above features reads both version 1 and 2 HFiles, but only writes version 2 HFiles. A version 2 HFile is structured as follows:
<inlinemediaobject>
<imageobject>
<imagedata align="middle" valign="middle" fileref="hfilev2.png" />
</imageobject>
<textobject>
<phrase>HFile Version 2</phrase>
</textobject>
<caption>
<para>HFile Version 2
</para>
</caption>
</inlinemediaobject>
</para>
</section>
<section><title>Unified version 2 block format</title>
<para>In the version 2 every block in the data section contains the following fields: </para>
<orderedlist>
<listitem>
<para>8 bytes: Block type, a sequence of bytes equivalent to version 1's "magic records". Supported block types are: </para>
<orderedlist>
<listitem>
<para>DATA – data blocks
</para>
</listitem>
<listitem>
<para>
LEAF_INDEX – leaf-level index blocks in a multi-level-block-index
</para>
</listitem>
<listitem>
<para>
BLOOM_CHUNK – Bloom filter chunks
</para>
</listitem>
<listitem>
<para>
META – meta blocks (not used for Bloom filters in version 2 anymore)
</para>
</listitem>
<listitem>
<para>
INTERMEDIATE_INDEX – intermediate-level index blocks in a multi-level blockindex
</para>
</listitem>
<listitem>
<para>
ROOT_INDEX – root>level index blocks in a multi>level block index
</para>
</listitem>
<listitem>
<para>
FILE_INFO – the “file info” block, a small key>value map of metadata
</para>
</listitem>
<listitem>
<para>
BLOOM_META – a Bloom filter metadata block in the load>on>open section
</para>
</listitem>
<listitem>
<para>
TRAILER – a fixed>size file trailer. As opposed to the above, this is not an
HFile v2 block but a fixed>size (for each HFile version) data structure
</para>
</listitem>
<listitem>
<para>
INDEX_V1 – this block type is only used for legacy HFile v1 block
</para>
</listitem>
</orderedlist>
</listitem>
<listitem>
<para>Compressed size of the block's data, not including the header (int).
</para>
<para>
Can be used for skipping the current data block when scanning HFile data.
</para>
</listitem>
<listitem>
<para>Uncompressed size of the block's data, not including the header (int)</para>
<para>
This is equal to the compressed size if the compression algorithm is NON
</para>
</listitem>
<listitem>
<para>File offset of the previous block of the same type (long)</para>
<para>
Can be used for seeking to the previous data/index block
</para>
</listitem>
<listitem>
<para>Compressed data (or uncompressed data if the compression algorithm is NONE).</para>
</listitem>
</orderedlist>
<para>The above format of blocks is used in the following HFile sections:</para>
<orderedlist>
<listitem>
<para>Scanned block section. The section is named so because it contains all data blocks that need to be read when an HFile is scanned sequentially.  Also contains leaf block index and Bloom chunk blocks. </para>
</listitem>
<listitem>
<para>Non-scanned block section. This section still contains unified-format v2 blocks but it does not have to be read when doing a sequential scan. This section contains “meta” blocks and intermediate-level index blocks.
</para>
</listitem>
</orderedlist>
<para>We are supporting “meta” blocks in version 2 the same way they were supported in version 1, even though we do not store Bloom filter data in these blocks anymore. </para></section>
<section><title> Block index in version 2</title>
<para>There are three types of block indexes in HFile version 2, stored in two different formats (root and non-root): </para>
<orderedlist>
<listitem>
<para>Data index — version 2 multi-level block index, consisting of:</para>
<orderedlist>
<listitem>
<para>
Version 2 root index, stored in the data block index section of the file
</para>
</listitem>
<listitem>
<para>
Optionally, version 2 intermediate levels, stored in the non%root format in the data index section of the file. Intermediate levels can only be present if leaf level blocks are present
</para>
</listitem>
<listitem>
<para>
Optionally, version 2 leaf levels, stored in the non%root format inline with data blocks
</para>
</listitem>
</orderedlist>
</listitem>
<listitem>
<para>Meta index — version 2 root index format only, stored in the meta index section of the file</para>
</listitem>
<listitem>
<para>Bloom index — version 2 root index format only, stored in the “load-on-open” section as part of Bloom filter metadata.</para>
</listitem>
</orderedlist></section>
<section><title>
Root block index format in version 2</title>
<para>This format applies to:</para>
<orderedlist>
<listitem>
<para>Root level of the version 2 data index</para>
</listitem>
<listitem>
<para>Entire meta and Bloom indexes in version 2, which are always single-level. </para>
</listitem>
</orderedlist>
<para>A version 2 root index block is a sequence of entries of the following format, similar to entries of a version 1 block index, but storing on-disk size instead of uncompressed size. </para>
<orderedlist>
<listitem>
<para>Offset (long) </para>
<para>
This offset may point to a data block or to a deeper>level index block.
</para>
</listitem>
<listitem>
<para>On-disk size (int) </para>
</listitem>
<listitem>
<para>Key (a serialized byte array stored using Bytes.writeByteArray) </para>
<orderedlist>
<listitem>
<para>Key (VInt)
</para>
</listitem>
<listitem>
<para>Key bytes
</para>
</listitem>
</orderedlist>
</listitem>
</orderedlist>
<para>A single-level version 2 block index consists of just a single root index block. To read a root index block of version 2, one needs to know the number of entries. For the data index and the meta index the number of entries is stored in the trailer, and for the Bloom index it is stored in the compound Bloom filter metadata.</para>
<para>For a multi-level block index we also store the following fields in the root index block in the load-on-open section of the HFile, in addition to the data structure described above:</para>
<orderedlist>
<listitem>
<para>Middle leaf index block offset</para>
</listitem>
<listitem>
<para>Middle leaf block on-disk size (meaning the leaf index block containing the reference to the “middle” data block of the file) </para>
</listitem>
<listitem>
<para>The index of the mid-key (defined below) in the middle leaf-level block.</para>
</listitem>
</orderedlist>
<para/>
<para>These additional fields are used to efficiently retrieve the mid-key of the HFile used in HFile splits, which we define as the first key of the block with a zero-based index of (n – 1) / 2, if the total number of blocks in the HFile is n. This definition is consistent with how the mid-key was determined in HFile version 1, and is reasonable in general, because blocks are likely to be the same size on average, but we don’t have any estimates on individual key/value pair sizes. </para>
<para/>
<para>When writing a version 2 HFile, the total number of data blocks pointed to by every leaf-level index block is kept track of. When we finish writing and the total number of leaf-level blocks is determined, it is clear which leaf-level block contains the mid-key, and the fields listed above are computed.  When reading the HFile and the mid-key is requested, we retrieve the middle leaf index block (potentially from the block cache) and get the mid-key value from the appropriate position inside that leaf block.</para></section>
<section><title>
Non-root block index format in version 2</title>
<para>This format applies to intermediate-level and leaf index blocks of a version 2 multi-level data block index. Every non-root index block is structured as follows. </para>
<orderedlist>
<listitem>
<para>numEntries: the number of entries (int). </para>
</listitem>
<listitem>
<para>entryOffsets: the “secondary index” of offsets of entries in the block, to facilitate a quick binary search on the key (numEntries + 1 int values). The last value is the total length of all entries in this index block. For example, in a non-root index block with entry sizes 60, 80, 50 the “secondary index” will contain the following int array: {0, 60, 140, 190}.</para>
</listitem>
<listitem>
<para>Entries. Each entry contains: </para>
<orderedlist>
<listitem>
<para>
Offset of the block referenced by this entry in the file (long)
</para>
</listitem>
<listitem>
<para>
On>disk size of the referenced block (int)
</para>
</listitem>
<listitem>
<para>
Key. The length can be calculated from entryOffsets.
</para>
</listitem>
</orderedlist>
</listitem>
</orderedlist></section><section><title>
Bloom filters in version 2</title>
<para>In contrast with version 1, in a version 2 HFile Bloom filter metadata is stored in the load-on-open section of the HFile for quick startup. </para>
<orderedlist>
<listitem>
<para>A compound Bloom filter. </para>
<orderedlist>
<listitem>
<para>
Bloom filter version = 3 (int). There used to be a DynamicByteBloomFilter class that had the Bloom filter version number 2
</para>
</listitem>
<listitem>
<para>
The total byte size of all compound Bloom filter chunks (long)
</para>
</listitem>
<listitem>
<para>
Number of hash functions (int
</para>
</listitem>
<listitem>
<para>
Type of hash functions (int)
</para>
</listitem>
<listitem>
<para>
The total key count inserted into the Bloom filter (long)
</para>
</listitem>
<listitem>
<para>
The maximum total number of keys in the Bloom filter (long)
</para>
</listitem>
<listitem>
<para>
The number of chunks (int)
</para>
</listitem>
<listitem>
<para>
Comparator class used for Bloom filter keys, a UTF>8 encoded string stored using Bytes.writeByteArray
</para>
</listitem>
<listitem>
<para>
Bloom block index in the version 2 root block index format
</para>
</listitem>
</orderedlist>
</listitem>
</orderedlist></section><section><title>File Info format in versions 1 and 2</title>
<para>The file info block is a serialized <ulink url="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/io/HbaseMapWritable.html">HbaseMapWritable</ulink> (essentially a map from byte arrays to byte arrays) with the following keys, among others. StoreFile-level logic adds more keys to this.</para>
<informaltable frame="all">
<tgroup cols="2"><tbody><row>
<entry>
<para>hfile.LASTKEY </para>
</entry>
<entry>
<para>The last key of the file (byte array) </para>
</entry>
</row>
<row>
<entry>
<para>hfile.AVG_KEY_LEN </para>
</entry>
<entry>
<para>The average key length in the file (int) </para>
</entry>
</row>
<row>
<entry>
<para>hfile.AVG_VALUE_LEN </para>
</entry>
<entry>
<para>The average value length in the file (int) </para>
</entry>
</row></tbody></tgroup>
</informaltable>
<para>File info format did not change in version 2. However, we moved the file info to the final section of the file, which can be loaded as one block at the time the HFile is being opened. Also, we do not store comparator in the version 2 file info anymore. Instead, we store it in the fixed file trailer. This is because we need to know the comparator at the time of parsing the load-on-open section of the HFile.</para></section><section><title>
Fixed file trailer format differences between versions 1 and 2</title>
<para>The following table shows common and different fields between fixed file trailers in versions 1 and 2. Note that the size of the trailer is different depending on the version, so it is “fixed” only within one version. However, the version is always stored as the last four-byte integer in the file. </para>
<para/>
<informaltable frame="all">
<tgroup cols="2">
<colspec colname='c1'/>
<colspec colname='c2'/>
<tbody>
<row>
<entry>
<para>Version 1 </para>
</entry>
<entry>
<para>Version 2 </para>
</entry>
</row>
<row>
<entry align="center" namest="c1" nameend="c2">
<para>File info offset (long) </para>
</entry>
</row>
<row>
<entry>
<para>Data index offset (long) </para>
</entry>
<entry>
<para>loadOnOpenOffset (long)</para>
<para><emphasis>The offset of the section that we need toload when opening the file.</emphasis></para>
</entry>
</row>
<row>
<entry align="center" namest="c1" nameend="c2">
<para>Number of data index entries (int) </para>
</entry>
</row>
<row>
<entry>
<para>metaIndexOffset (long)</para>
<para>This field is not being used by the version 1 reader, so we removed it from version 2.</para>
</entry>
<entry>
<para>uncompressedDataIndexSize (long)</para>
<para>The total uncompressed size of the whole data block index, including root-level, intermediate-level, and leaf-level blocks.</para>
</entry>
</row>
<row>
<entry namest="c1" nameend="c2" align="center">
<para>Number of meta index entries (int) </para>
</entry>
</row>
<row>
<entry namest="c1" nameend="c2" align="center">
<para>Total uncompressed bytes (long) </para>
</entry>
</row>
<row>
<entry>
<para>numEntries (int) </para>
</entry>
<entry>
<para>numEntries (long) </para>
</entry>
</row>
<row>
<entry namest="c1" nameend="c2" align="center">
<para>Compression codec: 0 = LZO, 1 = GZ, 2 = NONE (int) </para>
</entry>
</row>
<row>
<entry>
<para></para>
</entry>
<entry>
<para>The number of levels in the data block index (int) </para>
</entry>
</row>
<row>
<entry>
<para></para>
</entry>
<entry>
<para>firstDataBlockOffset (long)</para>
<para>The offset of the first first data block. Used when scanning. </para>
</entry>
</row>
<row>
<entry>
<para></para>
</entry>
<entry>
<para>lastDataBlockEnd (long)</para>
<para>The offset of the first byte after the last key/value data block. We don't need to go beyond this offset when scanning. </para>
</entry>
</row>
<row>
<entry>
<para>Version: 1 (int) </para>
</entry>
<entry>
<para>Version: 2 (int) </para>
</entry>
</row></tbody></tgroup>
</informaltable>
<para/></section>
<section><title>getShortMidpointKey(an optimization for data index block)</title>
<para>Note: this optimization was introduced in HBase 0.95+</para>
<para>HFiles contain many blocks that contain a range of sorted Cells. Each cell has a key. To save IO when reading Cells, the HFile also has an index that maps a Cell's start key to the offset of the beginning of a particular block. Prior to this optimization, HBase would use the key of the first cell in each data block as the index key.</para>
<para>In HBASE-7845, we generate a new key that is lexicographically larger than the last key of the previous block and lexicographically equal or smaller than the start key of the current block. While actual keys can potentially be very long, this "fake key" or "virtual key" can be much shorter. For example, if the stop key of previous block is "the quick brown fox", the start key of current block is "the who", we could use "the r" as our virtual key in our hfile index.</para>
<para>There are two benefits to this:</para>
<itemizedlist>
<listitem><section>having shorter keys reduces the hfile index size, (allowing us to keep more indexes in memory), and</section></listitem>
<listitem><section>using something closer to the end key of the previous block allows us to avoid a potential extra IO when the target key lives in between the "virtual key" and the key of the first element in the target block.</section></listitem>
</itemizedlist>
<para>This optimization (implemented by the getShortMidpointKey method) is inspired by LevelDB's ByteWiseComparatorImpl::FindShortestSeparator() and FindShortSuccessor().</para>
</section></section></appendix>
<appendix xml:id="other.info">
<title>Other Information About HBase</title>
<section xml:id="other.info.videos"><title>HBase Videos</title>
<para>Introduction to HBase
<itemizedlist>
<listitem><link xlink:href="http://www.cloudera.com/content/cloudera/en/resources/library/presentation/chicago_data_summit_apache_hbase_an_introduction_todd_lipcon.html">Introduction to HBase</link> by Todd Lipcon (Chicago Data Summit 2011).
</listitem>
<listitem><link xlink:href="http://www.cloudera.com/videos/intorduction-hbase-todd-lipcon">Introduction to HBase</link> by Todd Lipcon (2010).
</listitem>
</itemizedlist>
</para>
<para><link xlink:href="http://www.cloudera.com/videos/hadoop-world-2011-presentation-video-building-realtime-big-data-services-at-facebook-with-hadoop-and-hbase">Building Real Time Services at Facebook with HBase</link> by Jonathan Gray (Hadoop World 2011).
</para>
<para><link xlink:href="http://www.cloudera.com/videos/hw10_video_how_stumbleupon_built_and_advertising_platform_using_hbase_and_hadoop">HBase and Hadoop, Mixing Real-Time and Batch Processing at StumbleUpon</link> by JD Cryans (Hadoop World 2010).
</para>
</section>
<section xml:id="other.info.pres"><title>HBase Presentations (Slides)</title>
<para><link xlink:href="http://www.cloudera.com/content/cloudera/en/resources/library/hadoopworld/hadoop-world-2011-presentation-video-advanced-hbase-schema-design.html">Advanced HBase Schema Design</link> by Lars George (Hadoop World 2011).
</para>
<para><link xlink:href="http://www.slideshare.net/cloudera/chicago-data-summit-apache-hbase-an-introduction">Introduction to HBase</link> by Todd Lipcon (Chicago Data Summit 2011).
</para>
<para><link xlink:href="http://www.slideshare.net/cloudera/hw09-practical-h-base-getting-the-most-from-your-h-base-install">Getting The Most From Your HBase Install</link> by Ryan Rawson, Jonathan Gray (Hadoop World 2009).
</para>
</section>
<section xml:id="other.info.papers"><title>HBase Papers</title>
<para><link xlink:href="http://research.google.com/archive/bigtable.html">BigTable</link> by Google (2006).
</para>
<para><link xlink:href="http://www.larsgeorge.com/2010/05/hbase-file-locality-in-hdfs.html">HBase and HDFS Locality</link> by Lars George (2010).
</para>
<para><link xlink:href="http://ianvarley.com/UT/MR/Varley_MastersReport_Full_2009-08-07.pdf">No Relation: The Mixed Blessings of Non-Relational Databases</link> by Ian Varley (2009).
</para>
</section>
<section xml:id="other.info.sites"><title>HBase Sites</title>
<para><link xlink:href="http://www.cloudera.com/blog/category/hbase/">Cloudera's HBase Blog</link> has a lot of links to useful HBase information.
<itemizedlist>
<listitem><link xlink:href="http://www.cloudera.com/blog/2010/04/cap-confusion-problems-with-partition-tolerance/">CAP Confusion</link> is a relevant entry for background information on
distributed storage systems.
</listitem>
</itemizedlist>
</para>
<para><link xlink:href="http://wiki.apache.org/hadoop/HBase/HBasePresentations">HBase Wiki</link> has a page with a number of presentations.
</para>
<para><link xlink:href="http://refcardz.dzone.com/refcardz/hbase">HBase RefCard</link> from DZone.
</para>
</section>
<section xml:id="other.info.books"><title>HBase Books</title>
<para><link xlink:href="http://shop.oreilly.com/product/0636920014348.do">HBase: The Definitive Guide</link> by Lars George.
</para>
</section>
<section xml:id="other.info.books.hadoop"><title>Hadoop Books</title>
<para><link xlink:href="http://shop.oreilly.com/product/9780596521981.do">Hadoop: The Definitive Guide</link> by Tom White.
</para>
</section>
</appendix>
<appendix xml:id="hbase.history"><title>HBase History</title>
<itemizedlist>
<listitem>2006: <link xlink:href="http://research.google.com/archive/bigtable.html">BigTable</link> paper published by Google.
</listitem>
<listitem>2006 (end of year): HBase development starts.
</listitem>
<listitem>2008: HBase becomes Hadoop sub-project.
</listitem>
<listitem>2010: HBase becomes Apache top-level project.
</listitem>
</itemizedlist>
</appendix>
<appendix xml:id="asf" ><title>HBase and the Apache Software Foundation</title>
<para>HBase is a project in the Apache Software Foundation and as such there are responsibilities to the ASF to ensure
a healthy project.</para>
<section xml:id="asf.devprocess"><title>ASF Development Process</title>
<para>See the <link xlink:href="http://www.apache.org/dev/#committers">Apache Development Process page</link>
for all sorts of information on how the ASF is structured (e.g., PMC, committers, contributors), to tips on contributing
and getting involved, and how open-source works at ASF.
</para>
</section>
<section xml:id="asf.reporting"><title>ASF Board Reporting</title>
<para>Once a quarter, each project in the ASF portfolio submits a report to the ASF board. This is done by the HBase project
lead and the committers. See <link xlink:href="http://www.apache.org/foundation/board/reporting">ASF board reporting</link> for more information.
</para>
</section>
</appendix>
<xi:include xmlns:xi="http://www.w3.org/2001/XInclude" href="tracing.xml" />
<xi:include xmlns:xi="http://www.w3.org/2001/XInclude" href="rpc.xml" />
<index xml:id="book_index">
<title>Index</title>
</index>
</book>