You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@nifi.apache.org by GitBox <gi...@apache.org> on 2021/02/16 19:21:14 UTC
[GitHub] [nifi] markap14 commented on pull request #4818: NIFI-7646, NIFI-8222: Reduced lock contention when updating Provenance Repository by buffering up to 1 MB of serialized records in memory; keep InputStream from content repo open when possible for reading across multiple FlowFiles in a session
markap14 commented on pull request #4818:
URL: https://github.com/apache/nifi/pull/4818#issuecomment-780061865
I created a few different flows to measure the performance of NiFi on the main branch vs. this branch to verify that these changes resulted in significant performance improvements. Here are the results:
```
GenerateFlowFile (0 bytes) -> UpdateAttribute
Branch | FlowFiles / 5 min | CPU Utilization
-----------------------------------------------
main | 14.4 million | 600-650%
this branch | 21 million | 600-650%
46% higher throughput. 0% more CPU used.
GenerateFlowFile (37 bytes JSON) -> ConvertRecord (JSON In, JSON Out) -> UpdateAttribute
Branch | FlowFiles / 5 min | CPU Utilization
-----------------------------------------------
main | 7.6 million | 750-800%
this branch | 12.4 million | 900-950%
63% higher throughput. Took a good bit more CPU but made that CPU available for use by the processor.
GenerateFlowFile (1024 bytes) -> MergeContent (binary concat, 1024 FlowFiles/bin) -> UpdateAttribute
Branch | FlowFiles / 5 min | CPU Utilization
-----------------------------------------------
main | 7.3 million | 600-650%
this branch | 16.2 million | 650-700%
122% higher throughput. Maybe 1/2 core more CPU used.
```
----------------------------------------------------------------
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:
users@infra.apache.org