[GitHub] franz1981 commented on issue #2494: ARTEMIS-2224 Reduce contention on LivePageCacheImpl

classic Classic list List threaded Threaded
1 message Options
Reply | Threaded
Open this post in threaded view
|

[GitHub] franz1981 commented on issue #2494: ARTEMIS-2224 Reduce contention on LivePageCacheImpl

GitBox
franz1981 commented on issue #2494: ARTEMIS-2224 Reduce contention on LivePageCacheImpl
URL: https://github.com/apache/activemq-artemis/pull/2494#issuecomment-454338688
 
 
   > @franz1981 thx for recommending this tool, much easier to use than jfr!
   
   I was sure you would have like it: is indeed "pragmatically better" for devs IMHO: just remember to enable debug non safepoints, but with JFR is the same, reading their docs!
   
   @qihongxu
   > At the end of test the perf boosted to 30k tps
   
   The current implementation (lock-free) of the live page cache suffer of the same performance limitation of a `LinkedList`: queries at the beginning and at the end of the list are faster ie `O(1)`, while in the middle are slower ie `O(n)`.
   To make things faster we need to exploit a better access pattern and provide a cursor-like implementation on it to be used on `QueueImpl::depage`: it should improve things at the point of having ~`O(1)` `getMessage` cost if the cursor will be used with a sequential access pattern, like an iterator.
   If you like the idea and it really seems to be important I can provide an implementaion for it ie 900M of page size and very small messages are not such a common use case TBH

----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
[hidden email]


With regards,
Apache Git Services