Thesharing commented on pull request #15229: URL: https://github.com/apache/flink/pull/15229#issuecomment-894141088
Thank you for this pull request, @zhuzhurk! In my opinion, there's only one thing that need to be taken care of. `DefaultScheduler#getAllocationsToReserve` will iterate over all vertices to get all allocations to reserve, whose computation is O(N). For batch jobs, every vertex is a pipelined region. So the computation may be O(N^2) when all vertices are scheduled at the beginning of the job. Would it be better if we maintain the state of allocation to reserve in `DefaultScheduler`? -- 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]
