Considerations with Presto on Amazon EMR

Consider the following limitations when you run Presto on Amazon EMR.

Presto command line executable

In Amazon EMR, PrestoDB and Trino both use the same command line executable, presto-cli, as in the following example.

presto-cli --catalog hive

Non-configurable Presto deployment properties

The version of Amazon EMR that you use determines the Presto deployment configurations that are available. For more information about these configuration properties, see Deploying Presto in the Presto documentation. The following table shows the different configuration options for Presto properties files.

File	Configurable
log.properties	PrestoDB: Configurable in Amazon EMR versions 4.0.0 and later. Use the presto-log configuration classification.
config.properties	PrestoDB: Configurable in Amazon EMR versions 4.0.0 and later. Use the presto-config configuration classification.
hive.properties	PrestoDB: Configurable in Amazon EMR versions 4.1.0 and later. Use the presto-connector-hive configuration classification.
node.properties	PrestoDB: Configurable in Amazon EMR version 5.6.0 and later. Use the presto-node configuration classification.
jvm.config	Not configurable.

PrestoDB installation

The application name Presto continues to be used to install PrestoDB on clusters.

You can install either PrestoDB or Trino, but you can't install both on a single cluster. If you specify both PrestoDB and Trino when you attempt to create a cluster, a validation error occurs and the cluster creation request fails.

EMRFS and PrestoS3FileSystem configuration

With Amazon EMR versions 5.12.0 and later, PrestoDB can use EMRFS. For more information, see EMR File System (EMRFS) in the Amazon EMR Management Guide. With earlier versions of Amazon EMR, PrestoS3FileSystem is the only configuration option.

You can use a security configuration to set up encryption for EMRFS data in Amazon S3. You can also use IAM roles for EMRFS requests to Amazon S3. For more information, see Understanding encryption options and Configure IAM roles for EMRFS requests to Amazon S3 in the Amazon EMR Management Guide.

Note

If you query underlying data in Amazon S3 with Amazon EMR version 5.12.0, Presto errors can occur. This is because Presto fails to pick up configuration classification values from emrfs-site.xml. As a workaround, create an emrfs subdirectory under usr/lib/presto/plugin/hive-hadoop2/ and create a symlink in usr/lib/presto/plugin/hive-hadoop2/emrfs to the existing /usr/share/aws/emr/emrfs/conf/emrfs-site.xml file. Then restart the presto-server process (sudo presto-server stop followed by sudo presto-server start).

You can override the EMRFS default and use the PrestoS3FileSystem instead. To do this, use the presto-connector-hive configuration classification to set hive.s3-file-system-type to PRESTO as shown in the following example. For more information, see Configure applications.

[
   {
      "Classification": "presto-connector-hive",
      "Properties": {
         "hive.s3-file-system-type": "PRESTO"
      }
   }
]

If you use PrestoS3FileSystem, use the presto-connector-hive configuration classification to configure PrestoS3FileSystem properties. For more information about available properties, see Amazon S3 configuration in the Hive Connector section of the Presto documentation. These settings do not apply to EMRFS.

Default setting for end user impersonation

By default, Amazon EMR versions 5.12.0 and later enable end user impersonation for access to HDFS. For more information, see End user impersonation in the Presto documentation. To change this setting with the presto-config configuration classification, set the hive.hdfs.impersonation.enabled property to false.

Default port for Presto web interface

By default, Amazon EMR configures the Presto web interface on the Presto coordinator to use port 8889 (for PrestoDB and Trino). To change the port, use the presto-config configuration classification to set the http-server.http.port property. For more information, see Config properties in the Deploying Presto section of Presto Documentation.

Issue with Hive Bucket execution in some releases

Presto version 152.3 has an issue with Hive bucket execution that causes significantly slower Presto query performance in some circumstances. Amazon EMR versions 5.0.3, 5.1.0, and 5.2.0 include this version of Presto. To mitigate this issue, use the presto-connector-hive configuration classification to set the hive.bucket-execution property to false, as shown in the following example.

[
   {
      "Classification": "presto-connector-hive",
      "Properties": {
         "hive.bucket-execution": "false"
      }
   }
]