[SERVER-24483] Consider utilising a pool based memory allocator Created: 09/Jun/16  Updated: 06/Dec/22  Resolved: 19/Jul/16

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

Type: Improvement Priority: Major - P3
Reporter: Alexander Gorrod Assignee: Backlog - Storage Execution Team
Resolution: Won't Fix Votes: 0
Labels: None
Remaining Estimate: Not Specified
Time Spent: Not Specified
Original Estimate: Not Specified

Issue Links:
Depends
depends on SERVER-24268 Investigate jemalloc as alternative t... Closed
Assigned Teams:
Storage Execution
Participants:

 Description   

There are known issues with memory fragmentation in the MongoDB 3.2 release when running certain workloads (tracked in SERVER-20306).

One way we may be able to reduce the amount of fragmentation would be to use separate allocator pools for different allocations. Especially within WiredTiger, where there is some knowledge about the expected lifespan of different allocations.



 Comments   
Comment by Alexander Gorrod [ 19/Jul/16 ]

The analysis of JEMalloc didn't deliver enough evidence to switch default allocators in MongoDB. Since tcmalloc doesn't support a pool based allocation scheme, we can't pursue this method of reducing allocator fragmentation for now.

Generated at Thu Feb 08 04:06:29 UTC 2024 using Jira 9.7.1#970001-sha1:2222b88b221c4928ef0de3161136cc90c8356a66.