[Asterisk-code-review] res sorcery memory cache: Add basic module implementation. (asterisk[master])

Matt Jordan asteriskteam at digium.com
Thu May 21 21:30:38 CDT 2015


Matt Jordan has posted comments on this change.

Change subject: res_sorcery_memory_cache: Add basic module implementation.
......................................................................


Patch Set 4:

Before I review this change further, I'd like to explore the purpose of the threadpool a bit more.

One of the drivers for this feature/improvement is the discovery of substantial contention on a database with a realtime sorcery backend. That contention is actually exacerbated by usage of a threadpool in the PJSIP stack: a single point of contention ends up causing the multi-threading to be less efficient than if we simply had a single thread.

So, the question is: what is the intended purpose of the threadpool? And if we can expect those threads to be doing a similar task, on a shared resource, then do we really need a threadpool in the first place?

Note that if we _can_ remove the threadpool, then we no longer need the configuration file here, which simplifies the feature and its inclusion. It also decreases the number of idle threads that sit around in Asterisk, which - given the amount of context switching we already have to do between Stasis threadpools, PJSIP threadpools, IAX2 threadpools, and channel threads, is kind of a good thing.

-- 
To view, visit https://gerrit.asterisk.org/480
To unsubscribe, visit https://gerrit.asterisk.org/settings

Gerrit-MessageType: comment
Gerrit-Change-Id: Id8a6a357ef5a83d466f81eee56a67d13eeb118b9
Gerrit-PatchSet: 4
Gerrit-Project: asterisk
Gerrit-Branch: master
Gerrit-Owner: Joshua Colp <jcolp at digium.com>
Gerrit-Reviewer: Joshua Colp <jcolp at digium.com>
Gerrit-Reviewer: Mark Michelson <mmichelson at digium.com>
Gerrit-Reviewer: Matt Jordan <mjordan at digium.com>
Gerrit-HasComments: No



More information about the asterisk-code-review mailing list