ENABLED_RUNTIME_FILTER_TYPES Query Option (Impala 4.0 or higher only)

The ENABLED_RUNTIME_FILTER_TYPES query option sets enabled runtime filter types to be applied to scanners. This option only applies to HDFS scan nodes and Kudu scan nodes. The following types are supported. Specify the enabled types by a comma-separated list of the following values or enable all types by "ALL".
  • BLOOM
  • MIN_MAX
  • IN_LIST

Default: "BLOOM,MIN_MAX"

Type: string

Added in: Impala 4.0

Usage notes:

Impala backend expects comma separated values to be in quotes when executing the SET statement. This is usually the case when running SET statement like SET ENABLED_RUNTIME_FILTER_TYPES="value1,value2" using a JDBC driver. When using Impala-shell client, the SET statement is not executed immediately but query options are updated in the client and applied as part of the following statement, so no quotes are required for Impala-shell. That is, we use SET statement like SET ENABLED_RUNTIME_FILTER_TYPES=value1,value2 when submitting the query to Impala backend via Impala-shell client.

Depending on the scan node type, Planner can schedule compatible runtime filter types as follows.
  • Kudu scan: BLOOM, MIN_MAX
  • HDFS scan on Parquet files: BLOOM, MIN_MAX
  • HDFS scan on ORC files: BLOOM, IN_LIST
  • HDFS scan on other kinds of files: BLOOM

Related information:

Runtime Filtering for Impala Queries (Impala 2.5 or higher only), RUNTIME_FILTER_MODE Query Option (Impala 2.5 or higher only)