Prestosql guide

This tutorial provides a quick introduction to using current integration/presto module.

Presto Multinode Cluster Setup for Carbondata

Presto Single Node Setup for Carbondata

Presto Setup with CarbonData Distributed IndexServer

Presto SQL is now Trino

⚠️ PrestoSQL is rename as Trino. The software and the community you have come to love and depend on aren’t going anywhere, we are simply renaming. Trino is the new name for PrestoSQL, the project supported by the founders and creators of Presto® along with the major contributors – just under a shiny new name. And now you can find us here:

GitHub: https://github.com/trinodb/trino.

Presto Multinode Cluster Setup for Carbondata

Installing Presto

To know about which version of presto is supported by this version of carbon, visit https://github.com/apache/carbondata/blob/master/pom.xml and look for <presto.version> inside prestosql profile.

Example: <presto.version>333</presto.version> This means current version of carbon supports presto 333 version.

Note: Currently carbondata supports only one version of presto, cannot handle multiple versions at same time. If user wish to use older version of presto, then need to use older version of carbon (other old branches, say branch-1.5 and check the supported presto version in it's pom.xml file in integration/presto/)

1. Download that version of Presto (say 333) using below command:

wget https://repo1.maven.org/maven2/io/prestosql/presto-server/333/presto-server-333.tar.gz

2. Extract Presto tar file: tar zxvf presto-server-333.tar.gz.

3. Download the Presto CLI of the same presto server version (say 333) for the coordinator and name it presto.

  wget https://repo1.maven.org/maven2/io/prestosql/presto-cli/333/presto-cli-333-executable.jar

  mv presto-cli-333-executable.jar presto

  chmod +x presto

Create Configuration Files

1. Create etc folder in presto-server-333 directory.

2. Create config.properties, jvm.config, log.properties, and node.properties files.

3. Install uuid to generate a node.id.

   sudo apt-get install uuid
   
   uuid
   

Contents of your node.properties file

node.environment=production
node.id=<generated uuid>
node.data-dir=/home/ubuntu/data

Contents of your jvm.config file

-server
-Xmx16G
-XX:+UseG1GC
-XX:G1HeapRegionSize=32M
-XX:+UseGCOverheadLimit
-XX:+ExplicitGCInvokesConcurrent
-XX:+HeapDumpOnOutOfMemoryError
-XX:OnOutOfMemoryError=kill -9 %p

Contents of your log.properties file

io.prestosql=INFO

The default minimum level is INFO. There are four levels: DEBUG, INFO, WARN and ERROR.

Coordinator Configurations

Contents of your config.properties

coordinator=true
node-scheduler.include-coordinator=false
http-server.http.port=8086
query.max-memory=5GB
query.max-total-memory-per-node=5GB
query.max-memory-per-node=3GB
memory.heap-headroom-per-node=1GB
discovery-server.enabled=true
discovery.uri=http://<coordinator_ip>:8086

The options node-scheduler.include-coordinator=false and coordinator=true indicate that the node is the coordinator and tells the coordinator not to do any of the computation work itself and to use the workers.

Note: We recommend setting query.max-memory-per-node to half of the JVM config max memory, though if your workload is highly concurrent, you may want to use a lower value for query.max-memory-per-node.

Also relation between below two configuration-properties should be like: If, query.max-memory-per-node=30GB Then, query.max-memory=<30GB * number of nodes>.

Worker Configurations

Contents of your config.properties

coordinator=false
http-server.http.port=8086
query.max-memory=5GB
query.max-memory-per-node=2GB
discovery.uri=http://<coordinator_ip>:8086

Note: jvm.config and node.properties files are same for all the nodes (worker + coordinator). All the nodes should have different node.id.

Catalog Configurations

1. Create a folder named catalog in etc directory of presto on all the nodes of the cluster including the coordinator.

Configuring Carbondata in Presto

1. Create a file named carbondata.properties in the catalog folder and set the required properties on all the nodes.
2. As carbondata connector extends hive connector all the configurations(including S3) is same as hive connector. Just replace the connector name in hive configuration and copy same to carbondata.properties connector.name = carbondata

Add Plugins

1. Create a directory named carbondata in plugin directory of presto.
2. Copy all the jars from ../integration/presto/target/carbondata-presto-X.Y.Z-SNAPSHOT to plugin/carbondata directory on all nodes.

Start Presto Server on all nodes

./presto-server-333/bin/launcher start

To run it as a background process.

./presto-server-333/bin/launcher run

To run it in foreground.

Start Presto CLI

To connect to carbondata catalog use the following command:

./presto --server <coordinator_ip>:8086 --catalog carbondata --schema <schema_name>

Execute the following command to ensure the workers are connected.

select * from system.runtime.nodes;

Now you can use the Presto CLI on the coordinator to query data sources in the catalog using the Presto workers.

Presto Single Node Setup for Carbondata

Config presto server

• Download presto server (333 is suggested and supported) : https://repo1.maven.org/maven2/io/prestosql/presto-server/
• Finish presto configuration following https://github.com/trinodb/trino/blob/333/presto-docs/src/main/sphinx/installation/deployment.rst. A configuration example:

config.properties

coordinator=true
node-scheduler.include-coordinator=true
http-server.http.port=8086
query.max-memory=5GB
query.max-total-memory-per-node=5GB
query.max-memory-per-node=3GB
memory.heap-headroom-per-node=1GB
discovery-server.enabled=true
discovery.uri=http://localhost:8086
task.max-worker-threads=4
optimizer.dictionary-aggregation=true
optimizer.optimize-hash-generation = false  

jvm.config

-server
-Xmx4G
-XX:+UseG1GC
-XX:G1HeapRegionSize=32M
-XX:+UseGCOverheadLimit
-XX:+ExplicitGCInvokesConcurrent
-XX:+HeapDumpOnOutOfMemoryError
-XX:OnOutOfMemoryError=kill -9 %p
-XX:+TraceClassLoading
-Dcarbon.properties.filepath=<path>/carbon.properties

carbon.properties.filepath property is used to set the carbon.properties file path and it is recommended to set otherwise some features may not work. Please check the above example.

log.properties

io.prestosql=DEBUG
io.prestosql.server.PluginManager=DEBUG

node.properties

node.environment=carbondata
node.id=ffffffff-ffff-ffff-ffff-ffffffffffff
node.data-dir=/Users/apple/DEMO/presto_test/data

• Config carbondata-connector for presto

  Firstly: Compile carbondata, including carbondata-presto integration module

  $ git clone https://github.com/apache/carbondata
  $ cd carbondata
  $ mvn -DskipTests -P{spark-version} -P{prestodb/prestosql} clean package
  

  For prestosql-333 and above, please use spark2.3 profile. Because spark2.4 and higher profiles will bring hadoop3 dependencies and it will cause presto333 server launch failure. So, compile using:

  mvn -DskipTests -Pspark-2.3 -Pprestosql clean package
  

  Secondly: Create a folder named ‘carbondata’ under $PRESTO_HOME$/plugin and copy all jars from carbondata/integration/presto/target/carbondata-presto-x.x.x-SNAPSHOT to $PRESTO_HOME$/plugin/carbondata

  NOTE: Copying assemble jar alone will not work, need to copy all jars from integration/presto/target/carbondata-presto-x.x.x-SNAPSHOT

  Thirdly: Create a carbondata.properties file under $PRESTO_HOME$/etc/catalog/ containing the following contents:

  connector.name=carbondata
  hive.metastore.uri=thrift://<host>:<port>
  

  Carbondata becomes one of the supported format of presto hive plugin, so the configurations and setup is similar to hive connector of presto. Please refer https://github.com/trinodb/trino/blob/333/presto-docs/src/main/sphinx/connector/hive.rst for more details.

  Note: Since carbon can work only with hive metastore, it is necessary that spark also connects to same metastore db for creating tables and updating tables. All the operations done on spark will be reflected in presto immediately. It is mandatory to create Carbon tables from spark using CarbonData 1.5.2 or greater version since input/output formats are updated in carbon table properly from this version.

Connecting to carbondata store on s3

• In case you want to query carbonstore on S3 using S3A api put following additional properties inside $PRESTO_HOME$/etc/catalog/carbondata.properties

   Required properties
  
   hive.s3.aws-access-key={value}
   hive.s3.aws-secret-key={value}
  
   Optional properties
  
   hive.s3.endpoint={value}
  

  Please refer https://github.com/trinodb/trino/blob/333/presto-docs/src/main/sphinx/connector/hive.rst for more details on S3 integration.

Generate CarbonData file

There are two ways to create the carbondata files to query from presto.

1. Please refer to quick start: https://github.com/apache/carbondata/blob/master/docs/quick-start-guide.md. Load data statement in Spark can be used to create carbondata tables and then you can easily find the created carbondata files.
2. Carbondata supports writing the transactional data to table using the insert command. The folder structure will be similar to what we have for spark. Table can be created from spark, then data can be inserted from presto and queried from presto.

Data can be inserted in two ways.

1. insert into target_carbon_table select values('a,b,c');
2. insert into target_carbon_table select * from source_table;

Note: Load Carbon properties are not yet supported. The insert will work with all the default configurations.

Query carbondata in CLI of presto

• Download presto cli client of version 333 : https://repo1.maven.org/maven2/io/prestosql/presto-cli/

• Start CLI:

  $ ./presto --server localhost:8086 --catalog carbondata --schema default
  

  Replace the hostname, port and schema name with your own.

Supported features of presto carbon

Presto carbon only supports reading the carbon table which is written by spark carbon or carbon SDK. Regarding complex datatypes- currently reading of only Array and Struct datatypes are supported, while Map datatype is not yet supported. During reading, it supports the non-distributed index like block index and bloom index. It doesn't support Materialized View as it needs query plan to be changed and presto does not allow it. Also, Presto carbon supports streaming segment read from streaming table created by spark.

Presto Setup with CarbonData Distributed IndexServer

Dependency jars

After copying all the jars from ../integration/presto/target/carbondata-presto-X.Y.Z-SNAPSHOT to plugin/carbondata directory on all nodes, ensure copying the following jars as well.

1. Copy ../integration/spark/target/carbondata-spark_X.Y.Z-SNAPSHOT.jar
2. Copy corresponding Spark dependency jars to the location.

Configure properties

Configure IndexServer configurations in carbon.properties file. Refer Configuring IndexServer for more info. Add -Dcarbon.properties.filepath=<path>/carbon.properties in jvm.config file.

Presto with IndexServer

Start distributed index server. Launch presto CLI and fire SELECT query and check if the corresponding job is fired in the index server application. Users can use spark to view the cache loaded by using show metacache command. Refer: MetaCacheDDL