Details
-
Improvement
-
Status: Open
-
Major
-
Resolution: Unresolved
-
None
-
None
-
None
Description
To get table volumes, FileSystem::getContentSummary is widely used in TajoMaster and QueryMaster. It is used even multiple times for each query lifecycle. But, This API causes lots of overhead, especially in S3 with partitioned tables. The overhead also occurs in HDFS too with large partitioned tables.
The main objective of this issue is to eliminate Filesystem::getContentSummary as many as possible. This API is widely used in many code points. So, it would be better to move forward this issue as an umbrella issue.
Attachments
Issue Links
- relates to
-
TAJO-2063 Refactor FileTablespace::commitOutputData.
- Resolved