Performance Improvements pre Summit
The list things discussed for performance problems that should help:
- A default MV for RED metrics chart. (Done)
- Require service and operations filter before any search for attributes (Done)
- Disallow 30 days & 14 days trace periods (Done).
- Make trace API call non blocking and load whichever (RED metrics or trace API) returns earlier (comment #2657 (comment 1774550053)) (Done) cc @drosse
- Fix incorrect span count with an indicator #2692 (closed) (Half-completed)
Concerns:
Both the trace list API call and the RED metrics API calls are too slow (RED metrics is acceptable for very small time windows but doesn't even complete for larger time windows). Discussed option for a fix involves the summary MV which @joe-shaw is working on. See discussion on #2657 (comment 1773254774) and #2618 (comment 1780069970).
Edited by Arun Sori