[ 
https://issues.apache.org/jira/browse/FLINK-15103?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16992396#comment-16992396
 ] 

Piotr Nowojski commented on FLINK-15103:
----------------------------------------

FLINK-15172 sound interesting.

{quote}
I think we have more network buffers mostly in local execution environment. 
{quote}
Can we double check that? That would be fine I think, but would still require a 
mention in the release notes.
{quote}
I cannot guarantee that they have exactly same amount of network memory as 
before (since we also replaced the container cut-off and replaced with 
something different)
{quote}
I think this can also be fine as long as:
1. We think new default values are better (as far as I understand we do believe 
that :) ) 
2. We understand what has changed in what scenarios and we documented those 
things in:
  a) release notes for people upgrading from 1.9
  b) somewhere in the documentation (some performance tuning section?), for new 
people

The a) point should be straightforward: you were using options A, B on setups 
like X, Y, you might expect Foo.
The b) if you want to achieve Foo, on setups like X, Y, please turn the knob 
named A, B

Just to be clear, this issue of allocating more network buffers in local 
environment could be omitted from b).

> Performance regression on 3.12.2019 in various benchmarks
> ---------------------------------------------------------
>
>                 Key: FLINK-15103
>                 URL: https://issues.apache.org/jira/browse/FLINK-15103
>             Project: Flink
>          Issue Type: Bug
>          Components: Benchmarks
>            Reporter: Piotr Nowojski
>            Priority: Blocker
>             Fix For: 1.10.0
>
>
> Various benchmarks show a performance regression that happened on December 
> 3rd:
> [arrayKeyBy (probably the most easily 
> visible)|http://codespeed.dak8s.net:8000/timeline/#/?exe=1&ben=arrayKeyBy&env=2&revs=200&equid=off&quarts=on&extr=on]
>  
> [tupleKeyBy|http://codespeed.dak8s.net:8000/timeline/#/?exe=1&ben=tupleKeyBy&env=2&revs=200&equid=off&quarts=on&extr=on]
>  
> [twoInputMapSink|http://codespeed.dak8s.net:8000/timeline/#/?exe=1&ben=twoInputMapSink&env=2&revs=200&equid=off&quarts=on&extr=on]
>  [globalWindow (small 
> one)|http://codespeed.dak8s.net:8000/timeline/#/?exe=1&ben=globalWindow&env=2&revs=200&equid=off&quarts=on&extr=on]
>  and possible others.
> Probably somewhere between those commits: -8403fd4- 2d67ee0..60b3f2f



--
This message was sent by Atlassian Jira
(v8.3.4#803005)

Reply via email to