You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@hbase.apache.org by "Lars Hofhansl (JIRA)" <ji...@apache.org> on 2014/03/21 06:10:44 UTC
[jira] [Commented] (HBASE-10805) Speed up KeyValueHeap.next() a bit
[ https://issues.apache.org/jira/browse/HBASE-10805?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13942797#comment-13942797 ]
Lars Hofhansl commented on HBASE-10805:
---------------------------------------
Also tried with 20m rows, 1 col each, every in cache and filtered at the server. Without patch takes 5.63s, with 5.19s.
> Speed up KeyValueHeap.next() a bit
> ----------------------------------
>
> Key: HBASE-10805
> URL: https://issues.apache.org/jira/browse/HBASE-10805
> Project: HBase
> Issue Type: Bug
> Reporter: Lars Hofhansl
> Fix For: 0.99.0, 0.94.19, 0.98.2, 0.96.3
>
> Attachments: 10805-trunk.txt
>
>
> See discussion on HBASE-9969.
> This was brought up by [~mcorgan]. Even though I didn't believe him first.
> {code}
> KeyValueScanner topScanner = this.heap.peek();
> if (topScanner == null ||
> this.comparator.compare(kvNext, topScanner.peek()) >= 0) {
> this.heap.add(this.current);
> this.current = pollRealKV();
> }
> {code}
> We already have the invariant everywhere this.current always has a real KV polled. So adding current back to the heap followed by pollRealKV() is a no-op if this.current is the only scanner anyway.
> This can be changed to:
> {code}
> ...
> if (topScanner != null &&
> this.comparator.compare(kvNext, topScanner.peek()) >= 0) {
> ...
> {code}
> With 20m rows, 5 cols each, everything in the cache, fully compacted - i.e. one HFile per store, a scan that filters everything at the server through all 100m KVs takes 15.1s without the change and 13.3s with it, so a 12% improvement.
--
This message was sent by Atlassian JIRA
(v6.2#6252)