[
https://issues.apache.org/jira/browse/FLINK-15103?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16990847#comment-16990847
]
Stephan Ewen commented on FLINK-15103:
--------------------------------------
Fully agree with the {{[hotfix]}} assessment, this was more a {{[didn't want to
bother to create a ticket]}} situation, which is absolutely not okay for
anything that changes default configuration values!
Could you create a blocker issue to document the changed memory configuration?
This is indeed sensitive and must be spelled out prominently.
While much should work better/more robust out of the box, the one downside is
that the available heap for the memory state backend decreases by default. We
concluded that it was the "lesser of evils", because it ultimately only a
performance issue on an "unsafe" backend which always existed (now shows a
little earlier by default) and users can be expected to look into config
options when they want to tune performance. On the other hand, we now gain
memory robustness for RocksDB, good out-of-the-box behavior for batch, and a
simpler configuration model, which was something users struggled with hard and
early in the process.
> 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)
