imay commented on issue #1726: CPU idle is still high when concurrent access is high URL: https://github.com/apache/incubator-doris/issues/1726#issuecomment-527759718 After looking into this case, I found some problems that will hurt performance. 1. There are too many unnecessary memory allocation and copy in OlapScanNode. I have improved this in #1742 2. There are too many memory pool create/delete operation in our code path. And I have reduced some by reusing MemPool in #1732 But it is not enough, I will create another issue to track this problem. 3. One problem is related with my case. In my case the query is like "select day, sum(pv) from table where day = 20190904 group by day", all the concurrent queries will finish second phase aggregation in one backend of 45 backends. This lead that one backend's CPU is completely consumed, but other backends has about 60% CPU. I will create another issue to track this.
---------------------------------------------------------------- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: [email protected] With regards, Apache Git Services --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
