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

[GitBox]

qihongxu commented on issue #2494: ARTEMIS-2224 Reduce contention on 
LivePageCacheImpl
URL: https://github.com/apache/activemq-artemis/pull/2494#issuecomment-455086939
 
 
   @franz1981 
   Here's the result of single and 400 consumers.
   With these settings applied: 
   max-size-bytes = 950MB, 
   page-size = 900MB, 
   Produced 2,000,000 msgs in a new queue (make it a ~700MB page file as 
‘livePage’ + 950MB in journal), then closed producer, started consumer.
   
   **single consumer: avg 1k tps**
   ![1 consumer 1k 
tps](https://user-images.githubusercontent.com/10609615/51303793-6afb6e00-1a71-11e9-9228-e15c48a66c59.png)
   
   **400 consumers: avg 5.3k tps**
   ![400 consumer 5 4k 
tps](https://user-images.githubusercontent.com/10609615/51303961-dd6c4e00-1a71-11e9-9c48-75b7fd81e342.png)
   
   Their flamegraphes are similar like this.
   
![image](https://user-images.githubusercontent.com/10609615/51304005-012f9400-1a72-11e9-98e9-44dedccb0f6d.png)
   
   As we can see `getChunkOf(index, lastIndex)` is most costy in both cases, 
and 400 consumers didn't perform better than 200. Also, we found that once most 
journal msgs have been consumed and all paged msgs loaded to memory, tps can 
rocket to 30k just like we have shown in other tests before.  
   

----------------------------------------------------------------
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:
us...@infra.apache.org


With regards,
Apache Git Services