Details
-
Task
-
Resolution: Incomplete
-
P2: Important
-
None
-
None
-
None
Description
Currently the table test_results_raw has grown to a size of ~6G rows.
While we want to keep (some of) the old data and query them, we should do this optimally, given that the old data is read-only and very compressible.
Possibilities:
- PARTITION tables based on workitem started_on timestamp, on a per-month basis
- Move archived tables to a TABLESPACE on a compressed filesystem
-
- check first if the transparent compression of IT storage is good enough; this would make things much simpler.
- Use slow storage devices for old data
- Delete non-integration/nightly entries