Details
-
Improvement
-
Status: Closed
-
Major
-
Resolution: Fixed
-
None
Description
This issue creates a query-level cache for HMS requests. The lifecycle of that cache is associated to the lifecycle of the query. This basically means that each unique request to certain HMS APIs should only be served once from HMS, while follow-up repetitive calls will be retrieved from cache. The initial implementation includes caching for 19 APIs.
This issue also extends existing local HS2 HMS cache implementation introduced in HIVE-23949 to support other requests (getTableColumnStatistics, getPartitionsByNames). In fact, implementation relies on some of the logic introduced in that JIRA since there are some commonalities.
Attachments
Issue Links
- is related to
-
HIVE-24183 Clean up local HS2 HMS cache code
- Closed
- relates to
-
HIVE-23949 Introduce caching layer in HS2 to accelerate query compilation
- Closed
- links to