[SERVER-17033] Improve performance for bulk insert into WT with under oplog back pressure Created: 24/Jan/15  Updated: 31/Aug/15  Resolved: 31/Aug/15

Status: Closed
Project: Core Server
Component/s: Storage, WiredTiger
Affects Version/s: 3.0.0-rc6
Fix Version/s: None

Type: Improvement Priority: Major - P3
Reporter: Asya Kamsky Assignee: Martin Bligh
Resolution: Duplicate Votes: 0
Labels: WTcc
Remaining Estimate: Not Specified
Time Spent: Not Specified
Original Estimate: Not Specified

Issue Links:
Duplicate
duplicates SERVER-19551 Keep "milestones" against WT oplog to... Closed
Related
related to SERVER-16919 Oplog can grow much larger than confi... Closed
Backwards Compatibility: Fully Compatible
Sprint: Quint Iteration 7, QuInt 8 08/28/15, Quint 9 09/18/15
Participants:

 Description   

Using YCSB inserting 30M documents with 24 threads I see a drop in performance between build
871ed2eec7a1bf2bed5bc8fc5f04a3eaa99d3fbd and
612f66c61c12b705fe3c5ed251d63525f1bf650a

Running a test to confirm exactly which commit it was.



 Comments   
Comment by Martin Bligh [ 31/Aug/15 ]

Will be resolved by SERVER-19551

Comment by Daniel Pasette (Inactive) [ 25/Jan/15 ]

We'll leave this ticket open to examine the impact of back-pressue introduced in SERVER-16919

Comment by Daniel Pasette (Inactive) [ 24/Jan/15 ]

There are other oplog performance improvements planned, but not for 3.0.0

Comment by Daniel Pasette (Inactive) [ 24/Jan/15 ]

The previous version was broken and allowing the oplog to grow unbounded. Mathias' patch keeps the oplog size within a narrow size threshold and applies back pressure, which is necessary.

Generated at Thu Feb 08 03:43:05 UTC 2024 using Jira 9.7.1#970001-sha1:2222b88b221c4928ef0de3161136cc90c8356a66.