docs/_docs/sql-reference/ddl.adoc

// 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.
= Data Definition Language (DDL)

:toclevels:

This page encompasses all data definition language (DDL) commands supported by Ignite.

== CREATE TABLE

Create a new table and an underlying cache.

[source,sql]
----
CREATE TABLE [IF NOT EXISTS] tableName (tableColumn [, tableColumn]...
[, PRIMARY KEY (columnName [,columnName]...)])
[WITH "paramName=paramValue [,paramName=paramValue]..."]

tableColumn := columnName columnType [DEFAULT defaultValue] [PRIMARY KEY]
----


Parameters:

* `tableName` - name of the table.
* `tableColumn` - name and type of a column to be created in the new table.
* `columnName` - name of a previously defined column.
* `DEFAULT` - specifies a default value for the column. Only constant values are accepted.
* `IF NOT EXISTS` - create the table only if a table with the same name does not exist.
* `PRIMARY KEY` - specifies a primary key for the table that can consist of a single column or multiple columns.
* `WITH` - accepts additional parameters not defined by ANSI-99 SQL:

** `TEMPLATE=<cache's template name>` - case-sensitive​ name of a link:configuring-caches/configuration-overview#cache-templates[cache template]. A template is an instance of the `CacheConfiguration` class registered by calling `Ignite.addCacheConfiguration()`. Use predefined `TEMPLATE=PARTITIONED` or `TEMPLATE=REPLICATED` templates to create the cache with the corresponding replication mode. The rest of the parameters will be those that are defined in the `CacheConfiguration` object. By default, `TEMPLATE=PARTITIONED` is used if the template is not specified explicitly.
** `BACKUPS=<number of backups>` - sets the number of link:configuring-caches/configuring-backups[partition backups]. If neither this nor the `TEMPLATE` parameter is set, then the cache is created with `0` backup copies.
** `ATOMICITY=<ATOMIC | TRANSACTIONAL | TRANSACTIONAL_SNAPSHOT>` - sets link:key-value-api/transactions[atomicity mode] for the underlying cache. If neither this nor the `TEMPLATE` parameter is set, then the cache is created with the `ATOMIC` mode enabled. If `TRANSACTIONAL_SNAPSHOT` is specified, the table will link:transactions/mvcc[support transactions].
** `WRITE_SYNCHRONIZATION_MODE=<PRIMARY_SYNC | FULL_SYNC | FULL_ASYNC>` -
sets the write synchronization mode for the underlying cache. If neither this nor the `TEMPLATE` parameter is set, then the cache is created with `FULL_SYNC` mode enabled.
** `CACHE_GROUP=<group name>` - specifies the link:configuring-caches/cache-groups[group name] the underlying cache belongs to.
** `AFFINITY_KEY=<affinity key column name>` - specifies an link:data-modeling/affinity-collocation[affinity key] name which is a column of the `PRIMARY KEY` constraint.
** `CACHE_NAME=<custom name of the new cache>` - the name of the underlying cache created by the command.
** `DATA_REGION=<existing data region name>` - name of the link:memory-configuration/data-regions[data region] where table entries should be stored. By default, Ignite stores all the data in a default region.
** `KEY_TYPE=<custom name of the key type>` - sets the name of the custom key type that is used from the key-value APIs in Ignite. The name should correspond to a Java, .NET, or C++ class, or it can be a random one if link:data-modeling/data-modeling#binary-object-format[BinaryObjects] is used instead of a custom class. The number of fields and their types in the custom key type has to correspond to the `PRIMARY KEY`. Refer to the <<Description>> section below for more details.
** `VALUE_TYPE=<custom name of the value type of the new cache>` - sets the name of a custom value type that is used from the key-value and other non-SQL APIs in Ignite. The name should correspond to a Java, .NET, or C++ class, or it can be a random one if
link:data-modeling/data-modeling#binary-object-format[BinaryObjects] is used instead of a custom class. The value type should include all the columns defined in the CREATE TABLE command except for those listed in the `PRIMARY KEY` constraint. Refer to the <<Description>> section below for more details.
** `WRAP_KEY=<true | false>` - this flag controls whether a _single column_ `PRIMARY KEY` should be wrapped in the link:data-modeling/data-modeling#binary-object-format[BinaryObjects] format or not. By default, this flag is set to false. This flag does not have any effect on the `PRIMARY KEY` with multiple columns; it always gets wrapped regardless of the value of the parameter.
** `WRAP_VALUE=<true | false>` - this flag controls whether a single column value of a primitive type should be wrapped in the link:data-modeling/data-modeling#binary-object-format[BinaryObjects] format or not. By default, this flag is set to true. This flag does not have any effect on the value with multiple columns; it always gets wrapped regardless of the value of the parameter. Set this parameter to false if you have a single column value and do not plan to add additional columns to the table. Note that once the parameter is set to false, you can't use the `ALTER TABLE ADD COLUMN` command for this specific table.

The CREATE TABLE command creates a new Ignite cache and defines a SQL table on top of it. The cache stores the data in the form of key-value pairs while the table allows processing the data with SQL queries.

The table will reside in the schema specified in the connection parameters. If no schema is specified, the PUBLIC schema will be used. See link:SQL/schemas[Schemas] for more information about schemas in Ignite.

Note that the CREATE TABLE operation is synchronous and blocks the execution of other DDL commands that are issued while CREATE TABLE is still in progress. The execution of DML commands is not affected and can be performed in parallel.

If you wish to access the data using the key-value APIs, then setting the `CACHE_NAME`, `KEY_TYPE`, and `VALUE_TYPE` parameters may be useful for the following reasons:

- When the CREATE TABLE command is executed, the name of the cache is generated with the following format- `SQL_{SCHEMA_NAME}_{TABLE}`. Use the CACHE_NAME parameter to override the default name.
- Additionally, the command creates two new binary types - for the key and value respectively. Ignite generates the names of the types randomly including a UUID string. This complicates the usage of these 'types' from a non-SQL API. Use KEY_TYPE and VALUE_TYPE to override the names with custom ones corresponding to your business model objects.

Read more about the database architecture on the link:SQL/sql-introduction[SQL Introduction] page.


Examples:

Create Person table:

[source,sql]
----
CREATE TABLE IF NOT EXISTS Person (
  id int,
  city_id int,
  name varchar,
  age int,
  company varchar,
  PRIMARY KEY (id, city_id)
) WITH "template=partitioned,backups=1,affinity_key=city_id, key_type=PersonKey, value_type=MyPerson";
----

Once the CREATE TABLE command gets executed, the following happens:

- A new distributed cache is created and named SQL_PUBLIC_PERSON. This cache stores objects of the `Person` type that corresponds to a specific Java, .NET, C++ class or BinaryObject. Furthermore, the key type (`PersonKey`) and value type (`MyPerson`) are defined explicitly assuming the data is to be processed by key-value and other non-SQL APIs.
- SQL table/schema with all the parameters will be defined.
- The data will be stored in the form of key-value pairs. The `PRIMARY KEY` columns will be used as the object's key; the rest of the columns will belong to the value.
- Distributed cache related parameters are passed in the `WITH` clause of the statement. If the `WITH` clause is omitted, then the cache will be created with default parameters set in the CacheConfiguration object.

The example below shows how to create the same table with `PRIMARY KEY` specified in the column definition and overrid some cache related parameters:

[source,sql]
----
CREATE TABLE Person (
  id int PRIMARY KEY,
  city_id int,
  name varchar,
  age int,
  company varchar
) WITH "atomicity=transactional,cachegroup=somegroup";
----


== ALTER TABLE

Modify the structure of an existing table.

[source,sql]
----
ALTER TABLE [IF EXISTS] tableName {alter_specification}

alter_specification:
    ADD [COLUMN] {[IF NOT EXISTS] tableColumn | (tableColumn [,...])}
  | DROP [COLUMN] {[IF EXISTS] columnName | (columnName [,...])}
  | {LOGGING | NOLOGGING}

tableColumn := columnName columnType
----

[NOTE]
====
[discrete]
=== Scope of ALTER TABLE
Presently, Ignite only supports addition and removal of columns.
====

Parameters:

- `tableName` - the name of the table.
- `tableColumn` - the name and type of the column to be added to the table.
- `columnName` - the name of the column to be added or removed.
- `IF EXISTS` - if applied to TABLE, do not throw an error if a table with the specified table name does not exist. If applied to COLUMN, do not throw an error if a column with the specified name does not exist.
- `IF NOT EXISTS` - do not throw an error if a column with the same name already exists.
- `LOGGING` - enable link:persistence/native-persistence#write-ahead-log[write-ahead logging] for the table. Write-ahead logging in enabled by default. The command is relevant only if Ignite persistence is used.
- `NOLOGGING` - disable write-ahead logging for the table. The command is relevant only if Ignite persistence is used.


`ALTER TABLE ADD` adds a new column or several columns to a previously created table. Once a column is added, it can be accessed using link:sql-reference/dml[DML commands] and indexed with the <<CREATE INDEX>> statement.

`ALTER TABLE DROP` removes an existing column or multiple columns from a table. Once a column is removed, it cannot be accessed within queries. Consider the following notes and limitations:

- The command does not remove actual data from the cluster which means that if the column 'name' is dropped, the value of the 'name' is still stored in the cluster. This limitation is to be addressed in the next releases.
- If the column was indexed, the index has to be dropped manually using the 'DROP INDEX' command.
- It is not possible to remove a column that is a primary key or a part of such a key.
- It is not possible to remove a column if it represents the whole value stored in the cluster. The limitation is relevant for primitive values.
Ignite stores data in the form of key-value pairs and all the new columns will belong to the value. It's not possible to change a set of columns of the key (`PRIMARY KEY`).

Both DDL and DML commands targeting the same table are blocked for a short time until `ALTER TABLE` is in progress.

Schema changes applied by this command are persisted on disk if link:persistence/native-persistence[Ignite persistence] is enabled. Thus, the changes can survive full cluster restarts.


Examples:

Add a column to the table:

[source,sql]
----
ALTER TABLE Person ADD COLUMN city varchar;
----


Add a new column to the table only if a column with the same name does not exist:

[source,sql]
----
ALTER TABLE City ADD COLUMN IF NOT EXISTS population int;
----


Add a column​ only if the table exists:

[source,sql]
----
ALTER TABLE IF EXISTS Missing ADD number long;
----


Add several columns to the table at once:


[source,sql]
----
ALTER TABLE Region ADD COLUMN (code varchar, gdp double);
----


Drop a column from the table:


[source,sql]
----
ALTER TABLE Person DROP COLUMN city;
----


Drop a column from the table only if a column with the same name does exist:


[source,sql]
----
ALTER TABLE Person DROP COLUMN IF EXISTS population;
----


Drop a column only if the table exists:


[source,sql]
----
ALTER TABLE IF EXISTS Person DROP COLUMN number;
----


Drop several columns from the table at once:


[source,sql]
----
ALTER TABLE Person DROP COLUMN (code, gdp);
----


Disable write-ahead logging:


[source,sql]
----
ALTER TABLE Person NOLOGGING
----


== DROP TABLE

The `DROP TABLE` command drops an existing table.
The underlying cache with all the data in it is destroyed, too.


[source,sql]
----
DROP TABLE [IF EXISTS] tableName
----

Parameters:

- `tableName` - the name of the table.
- `IF NOT EXISTS` - do not throw an error if a table with the same name does not exist.


Both DDL and DML commands targeting the same table are blocked while the `DROP TABLE` is in progress.
Once the table is dropped, all pending commands will fail with appropriate errors.

Schema changes applied by this command are persisted on disk if link:persistence/native-persistence[Ignite persistence] is enabled. Thus, the changes can survive full cluster restarts.

Examples:

Drop Person table if the one exists:

[source,sql]
----
DROP TABLE IF EXISTS "Person";
----

== CREATE INDEX

Create an index on the specified table.

[source,sql]
----
CREATE [SPATIAL] INDEX [[IF NOT EXISTS] indexName] ON tableName
    (columnName [ASC|DESC] [,...]) [(index_option [...])]

index_option := {INLINE_SIZE size | PARALLEL parallelism_level}
----

Parameters:

* `indexName` - the name of the index to be created.
* `ASC` - specifies ascending sort order (default).
* `DESC` - specifies descending sort order.
* `SPATIAL` - create the spatial index. Presently, only geometry types are supported.
* `IF NOT EXISTS` - do not throw an error if an index with the same name already exists. The database checks indexes' names only, and does not consider columns types or count.
* `index_option` - additional options for index creation:
** `INLINE_SIZE` - specifies index inline size in bytes. Depending on the size, Ignite will place the whole indexed value or a part of it directly into index pages, thus omitting extra calls to data pages and increasing queries' performance. Index inlining is enabled by default and the size is pre-calculated automatically based on the table structure. To disable inlining, set the size to 0 (not recommended). Refer to the link:SQL/sql-tuning#increasing-index-inline-size[Increasing Index Inline Size] section for more details.
** `PARALLEL` - specifies the number of threads to be used in parallel for index creation. The greater number is set, the faster the index is created and built. If the value exceeds the number of CPUs, then it will be decreased to the number of cores. If the parameter is not specified, then the number of threads is calculated as 25% of the CPU cores available.


`CREATE INDEX` creates a new index on the specified table. Regular indexes are stored in the internal B+tree data structures. The B+tree gets distributed across the cluster along with the actual data. A cluster node stores a part of the index for the data it owns.

If `CREATE INDEX` is executed in runtime on live data then the database will iterate over the specified columns synchronously indexing them. The rest of the DDL commands targeting the same table are blocked until CREATE INDEX is in progress. DML command execution is not affected and can be performed in parallel.

Schema changes applied by this command are persisted on disk if link:persistence/native-persistence[Ignite persistence] is enabled. Thus, the changes can survive full cluster restarts.



=== Indexes Tradeoffs
There are multiple things you should consider when choosing indexes for your application.

- Indexes are not free. They consume memory, and each index needs to be updated separately, thus the performance of write operations might drop if too many indexes are created. On top of that, if a lot of indexes are defined, the optimizer might make more mistakes by choosing the wrong index while building the execution plan.
+
WARNING: It is poor strategy to index everything.

- Indexes are just sorted data structures (B+tree). If you define an index for the fields (a,b,c) then the records will be sorted first by a, then by b and only then by c.
+
[NOTE]
====
[discrete]
=== Example of Sorted Index
[width="25%" cols="33l, 33l, 33l"]
|=====
| A | B | C
| 1 | 2 | 3
| 1 | 4 | 2
| 1 | 4 | 4
| 2 | 3 | 5
| 2 | 4 | 4
| 2 | 4 | 5
|=====

Any condition like `a = 1 and b > 3` can be viewed as a bounded range, both bounds can be quickly looked up in *log(N)* time, the result will be everything between.

The following conditions will be able to use the index:

- `a = ?`
- `a = ? and b = ?`
- `a = ? and b = ? and c = ?`

Condition `a = ? and c = ?` is no better than `a = ?` from the index point of view.
Obviously half-bounded ranges like `a > ?` can be used as well.
====

- Indexes on single fields are no better than group indexes on multiple fields starting with the same field (index on (a) is no better than (a,b,c)). Thus it is preferable to use group indexes.

- When `INLINE_SIZE` option is specified, indexes holds a prefix of field data in the B+tree pages. This improves search performance by doing less row data retrievals, however substantially increases size of the tree (with a moderate increase in tree height) and reduces data insertion and removal performance due to excessive page splits and merges. It's a good idea to consider page size when choosing inlining size for the tree: each B-tree entry requires `16 + inline-size` bytes in the page (plus header and extra links for the page).


Examples:

Create a regular index:

[source,sql]
----
CREATE INDEX title_idx ON books (title);
----

Create a descending index only if it does not exist:

[source,sql]
----
CREATE INDEX IF NOT EXISTS name_idx ON persons (firstName DESC);
----

Create a composite index:

[source,sql]
----
CREATE INDEX city_idx ON sales (country, city);
----

Create an index specifying data inline size:

[source,sql]
----
CREATE INDEX fast_city_idx ON sales (country, city) INLINE_SIZE 60;
----

Create a geospatial​ index:

[source,sql]
----
CREATE SPATIAL INDEX idx_person_address ON Person (address);
----


== DROP INDEX

`DROP INDEX` deletes an existing index.


[source,sql]
----
DROP INDEX [IF EXISTS] indexName
----

Parameters:

* `indexName` - the name of the index to drop.
* `IF EXISTS` - do not throw an error if an index with the specified name does not exist. The database checks indexes' names only not considering column types or count.


DDL commands targeting the same table are blocked until `DROP INDEX` is in progress. DML command execution is not affected and can be performed in parallel.

Schema changes applied by this command are persisted on disk if link:persistence/native-persistence[Ignite persistence] is enabled. Thus, the changes can survive full cluster restarts.


[discrete]
=== Examples
Drop an index:


[source,sql]
----
DROP INDEX idx_person_name;
----


== CREATE USER

The command creates a user with a given name and password.

A new user can only be created using a superuser account when authentication for thin clients is enabled. Ignite creates the superuser account under the name `ignite` and password `ignite` on the first cluster start-up. Presently, you can't rename the superuser account nor grant its privileges to any other account.



[source,sql]
----
CREATE USER userName WITH PASSWORD 'password';
----

Parameters:

* `userName` - new user's name. The name cannot be longer than 60 bytes in UTF8 encoding.
* `password` - new user's password. An empty password is not allowed.

To create a _case-sensitive_ username, use the quotation (") SQL identifier.

[NOTE]
====
[discrete]
=== When Are Case-Sensitive Names Preferred?
The case-insensitivity property of the usernames is supported for JDBC and ODBC interfaces only. If it's planned to access Ignite from Java, .NET, or other programming language APIs then the username has to be passed either in all upper-case letters or enclosed in double quotes (") from those interfaces.

For instance, if `Test` was set as a username then:

- You can use `Test`, `TEst`, `TEST` and other combinations from JDBC and ODBC.
- You can use either `TEST` or `"Test"` as the username from Ignite's native SQL APIs designed for Java, .NET and other programming languages.

Alternatively, use the case-sensitive username at all times to ensure name consistency across all the SQL interfaces.
====

Examples:

Create a new user using test as a name and password:


[source,sql]
----
CREATE USER test WITH PASSWORD 'test';
----

Create a case-sensitive username:


[source,sql]
----
CREATE USER "TeSt" WITH PASSWORD 'test'
----


== ALTER USER

The command changes an existing user's password.
The password can be updated by the superuser (`ignite`, see <<CREATE USER>> for more details) or by the user themselves.


[source,sql]
----
ALTER USER userName WITH PASSWORD 'newPassword';
----


Parameters:

* `userName` - existing user's name.
* `newPassword` - the new password to set for the user's account.


Examples:

Updating user's password:


[source,sql]
----
ALTER USER test WITH PASSWORD 'test123';
----


== DROP USER

The command removes an existing user.

The user can be removed only by the superuser (`ignite`, see <<CREATE USER>> for more details).


[source,sql]
----
DROP USER userName;
----


Parameters:

* `userName` - a name of the user to remove.


Examples:

[source,sql]
----
DROP USER test;
----