Details
-
Sub-task
-
Status: Resolved
-
Major
-
Resolution: Fixed
-
2.2.0
-
None
Description
If a table is handled by DruidStorageHandler, we should be able to recognize it in Calcite. This way we will be able to apply rules in the optimizer to push an operator pipeline into the DruidScan. This will help us generating Druid featured queries e.g. timeseries, topn, groupby, or select.
For that purpose, we need to introduce a DruidQuery operator based on TableScan. In particular, we should leverage the code introduced in CALCITE-1121. Other operators will be pushed to DruidQuery if they can be executed by Druid i.e. the decision will be made statically (if we can push computation to Druid, we will do it). However, in the future the decision should be cost-based.
Attachments
Issue Links
- is part of
-
HIVE-14217 Druid integration
- Closed