[SERVER-68960] Investigate performance of point index scans Created: 18/Aug/22 Updated: 27/Oct/23 Resolved: 28/Mar/23 |
|
| Status: | Closed |
| Project: | Core Server |
| Component/s: | None |
| Affects Version/s: | None |
| Fix Version/s: | None |
| Type: | Task | Priority: | Major - P3 |
| Reporter: | Mihai Andrei | Assignee: | Backlog - Query Execution |
| Resolution: | Gone away | Votes: | 0 |
| Labels: | pm2697-m4 | ||
| Remaining Estimate: | Not Specified | ||
| Time Spent: | Not Specified | ||
| Original Estimate: | Not Specified | ||
| Issue Links: |
|
||||||||
| Assigned Teams: |
Query Execution
|
||||||||
| Participants: | |||||||||
| Linked BF Score: | 10 | ||||||||
| Story Points: | 5 | ||||||||
| Description |
|
SERVER-66437 showed that IDHACK in SBE is slower than in classic largely due to the time that it takes to prepare a plan executor. This task looks into whether a similar regression exists for point index scans on non-id indexes (that is, over indexes that can't use IDHACK) |
| Comments |
| Comment by Mihai Andrei [ 28/Mar/23 ] |
|
NonIntIdFindOne has no more regressions |
| Comment by Kyle Suarez [ 27/Oct/22 ] |
|
Ah, nice. My vote is "gone away". |
| Comment by Anna Wawrzyniak [ 27/Oct/22 ] |
|
Non-id lookups suffer from the same preparation overhead as id lookups. I don't think there is much we can do to improve the perf here, other than solve the preparation problem (e.g. by pooling: https://jira.mongodb.org/browse/SERVER-69774 ). Should we just mark this ticket as dependent on SERVER-69774 ? |
| Comment by Kyle Suarez [ 21/Sep/22 ] |
|
anna.wawrzyniak@mongodb.com, we're assigning this to you since it's related to your work caching prepared plans in SERVER-69774. CC david.storch@mongodb.com, mihai.andrei@mongodb.com, andrew.paroski@mongodb.com |