msbukal commented on pull request #15873: URL: https://github.com/apache/beam/pull/15873#issuecomment-962118962
I ran some performance tests with an import of 68 million resources. With sharding, aka before this PR: | Elapsed Time | # Import Operations | Resources Imported | Results | | ------------- | -------------------- | -------------------- | ------- | | 1h33 | 3 | 9,803,961 | FAILED due to java.lang.IllegalStateException: Not all temporary files are present for importing | | 2h6 | 3 | 34,706,066 | FAILED due to java.lang.IllegalStateException: Not all temporary files are present for importing | | 2h37 | 16 | 67,480,522 | Finally got a successful run, only used 1 worker. | Without sharding, aka after this PR: | Elapsed Time | # Import Operations | Resources Imported | Results | | ------------- | -------------------- | -------------------- | ------- | | 1h30 | 12 | 67,480,522 | Scaled to ~6 workers. | | 2h17 | 4 | 67,480,522 | Force limited to 1 worker | Overall, without pre-sharding performed better on the two key points: Less operations triggered (good to avoid hitting quota), and less elapsed time.. additionally didn’t fail as constantly as with sharding which was more problematic than I expected. -- 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. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected]
