Uploaded image for project: 'Cassandra'
  1. Cassandra
  2. CASSANDRA-3855

RemoveDeleted dominates compaction time for large sstable counts

    XMLWordPrintableJSON

    Details

    • Severity:
      Normal

      Description

      With very large numbers of sstables (2000+ generated by a `bin/stress -n 100,000,000` run with LeveledCompactionStrategy), PrecompactedRow.removeDeletedAndOldShards dominates compaction runtime, such that commenting it out takes compaction throughput from 200KB/s to 12MB/s.

      Stack attached.

        Attachments

        1. with-cleaning-java.hprof.txt
          53 kB
          Stu Hood
        2. 3855.txt
          1 kB
          Yuki Morishita

          Activity

            People

            • Assignee:
              yukim Yuki Morishita
              Reporter:
              stuhood Stu Hood
              Authors:
              Yuki Morishita
              Reviewers:
              Jonathan Ellis
            • Votes:
              0 Vote for this issue
              Watchers:
              3 Start watching this issue

              Dates

              • Created:
                Updated:
                Resolved: