[FLINK-15103] Performance regression on 3.12.2019 in various benchmarks

• Type: Bug
• Status: Closed
• Priority: Blocker
• Resolution: Won't Do
• Fix Version/s: 1.10.0
• Component/s: Benchmarks
• Labels: pull-request-available
• Assignee: Piotr Nowojski
• Reporter: Piotr Nowojski
• Created: 06/Dec/19 12:21
• Updated: 06/Jan/20 11:01
• Resolved: 06/Jan/20 08:55

Description

Various benchmarks show a performance regression that happened on December 3rd:

arrayKeyBy (probably the most easily visible)
tupleKeyBy
twoInputMapSink
globalWindow (small one)

and possible others.

Probably somewhere between those commits: 8403fd4 2d67ee0..60b3f2f

Issue Links

• is caused by: FLINK-13980 FLIP-49 Unified Memory Configuration for TaskExecutors
• relates to: FLINK-15143 Create document for FLIP-49 TM memory model and configuration guide
• relates to: FLINK-15144 Document performance related changes caused by FLIP-49
• links to: GitHub Pull Request #37