[BEAM-7513] [SQL] Row Estimation for BigQueryTable

• Type: New Feature
• Status: Triage Needed
• Priority: P2
• Resolution: Implemented
• Fix Version/s: 2.14.0
• Component/s: dsl-sql, io-java-gcp
• Assignee: Alireza Samadianzakaria
• Reporter: Alireza Samadianzakaria
• Created: 07/Jun/19 23:21
• Updated: 13/Apr/23 10:56
• Resolved: 20/Jun/19 16:23

Description

Calcite tables (org.apache.calcite.schema.Table) should implement the method org.apache.calcite.schema.Statistic getStatistic(). The Statistic instance returned by this method is used for the Volcano optimizer in Calcite. 

Currently, org.apache.beam.sdk.extensions.sql.impl.BeamCalciteTable has not implemented getStatistic() which means it uses the implementation in org.apache.calcite.schema.impl.AbstractTable and that implementation just returns Statistics.UNKNOWN for all sources.

 

Things needed to be implemented:

1- Implementing getStatistic in BeamCalciteTable such that it calls a row count estimation method from BeamSqlTable and adding this method to BeamSqlTable.

2- Implementing the row count estimation method for BigQueryTable. 

Issue Links

• links to: GitHub Pull Request #8822
• links to: GitHub Pull Request #8882
• links to: GitHub Pull Request #8892