[ https://issues.apache.org/jira/browse/CASSANDRA-6662?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13898085#comment-13898085 ]
Aleksey Yeschenko edited comment on CASSANDRA-6662 at 2/11/14 6:06 PM: ----------------------------------------------------------------------- Accepted nits #1, #2, #4, #5 and committed, thanks. As for #3 - got rid of the binarySearch() there, but did leave the reconcile-with-previous-cell logic in place (for one, we've already done the comparison, so it makes sense to use the result; secondly, it does happen more often than you think - when merging results from memtabe+sstables on read in CollationController.collectAllData()). Actually, returned the binarySearch() logic back as well. It should have a significant enough effect on CollationController.collectTimeOrderedData(), and only adds one more branch on top, anyway (in 6de2fd9bfd4c20e47495d171da7175c1a9e92c68). was (Author: iamaleksey): Accepted nits #1, #2, #4 and #5 and committed, thanks. As for #3 - got rid of the binarySearch() there, but did leave the reconcile-with-previous-cell logic in place (for one, we've already done the comparison, so it makes sense to use the result; secondly, it does happen more often than you think - when merging results from memtabe+sstables on read in CollationController.collectAllData). Actually, returned the binarySearch() logic back as well. It should have a significant enough affect on CollationController.collectTimeOrderedData(), and only adds one more branch on top, anyway (in 6de2fd9bfd4c20e47495d171da7175c1a9e92c68). > Sort/reconcile cells in ArrayBackedSortedColumns only when an accessor is > called > -------------------------------------------------------------------------------- > > Key: CASSANDRA-6662 > URL: https://issues.apache.org/jira/browse/CASSANDRA-6662 > Project: Cassandra > Issue Type: Improvement > Reporter: Aleksey Yeschenko > Assignee: Aleksey Yeschenko > Fix For: 2.1 > > > To avoid poor performance with huge numbers of cells added out of order > (which should be rare, but *can* happen with certain batch scenarios) we > should make ABSC only sort/reconcile its cells when an actual accessor is > actually called, delaying sorting until the very end. -- This message was sent by Atlassian JIRA (v6.1.5#6160)