[SERVER-1528] with new index skipping, nscanned may not accurately predict time cost of a query Created: 31/Jul/10  Updated: 12/Jul/16  Resolved: 04/Aug/10

Status: Closed
Project: Core Server
Component/s: None
Affects Version/s: None
Fix Version/s: 1.7.0

Type: Bug Priority: Major - P3
Reporter: Aaron Staple Assignee: Aaron Staple
Resolution: Done Votes: 1
Labels: None
Remaining Estimate: Not Specified
Time Spent: Not Specified
Original Estimate: Not Specified

Issue Links:
Depends
Operating System: ALL
Participants:

 Description   

With the new btree skipping implementation, currently nscanned counts the number of documents which match the query according to the index spec. But this can result in situations in which indexes which require skipping have the same nscanned as indexes not requiring skipping.

One solution is to include in nscanned the number of btree entries touched during skipping. Currently there's no way to transfer that information from the btree implementation to the query optimizer, but that can be added.

I can implement, but sounds like a good idea Eliot?



 Comments   
Comment by auto [ 04/Aug/10 ]

Author:

{'login': 'astaple', 'name': 'Aaron', 'email': 'aaron@10gen.com'}

Message: SERVER-1528 account for expensive skipping operations in nscanned
http://github.com/mongodb/mongo/commit/701dc65282b7e154aca20ae1882ae7962c11e0b2

Comment by Eliot Horowitz (Inactive) [ 02/Aug/10 ]

going to make just 1.7 early for now.
if patch doesn't seem to bad - then can do earlier perhaps

Comment by Eliot Horowitz (Inactive) [ 31/Jul/10 ]

Yes but don't want to change that for 1.6.

Generated at Thu Feb 08 02:57:16 UTC 2024 using Jira 9.7.1#970001-sha1:2222b88b221c4928ef0de3161136cc90c8356a66.