zhijiangW commented on issue #11155: [FLINK-14818][benchmark] Fix receiving 
InputGate setup of StreamNetworkBenchmarkEnvironment.
URL: https://github.com/apache/flink/pull/11155#issuecomment-590720137
 
 
   Thanks for the updates @wsry .
   
   As for the issue of connection index, I prefer to keeping it as previous 
value, then we can only measure the performance effect caused by gate/channel 
amount.
   
   If we want to adjust the connection amount via `connectionIndex`, it should 
be a separate work to do in future if confirmed necessary.
   
   If the performance regression still exists after getting ride of connection 
factor, caused by total buffers reduction as you analyzed above, I am a bit 
torn whether we should increase the default floating buffer setting for these 
benchmarks in order to keep the performance same as before. Otherwise we might 
have a  bad/invalid curve to trace the performance changes in history term. 
WDYT? @pnowojski  

----------------------------------------------------------------
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

Reply via email to