[jira] [Commented] (HBASE-6066) some low hanging read path improvement ideas

["Kannan Muthukkaruppan (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/HBASE-6066?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13295779#comment-13295779 ] 

Kannan Muthukkaruppan commented on HBASE-6066:
----------------------------------------------

One of our interns (Aurick) is planning to work on this.

Part #3, the incrNumericMetric overhead patch is easier-- it is a slight variant of Todd's patch and should result in fewer AtomicLong operations because it just accumulates in a local variable and updates the counter at the end. It is part of a diff by Michael Chen: https://reviews.facebook.net/D3627#a83f6d09. I think it is better to fork off this part (#4) into its own JIRA and Aurick will work on the others.
                
> some low hanging read path improvement ideas 
> ---------------------------------------------
>
>                 Key: HBASE-6066
>                 URL: https://issues.apache.org/jira/browse/HBASE-6066
>             Project: HBase
>          Issue Type: Improvement
>            Reporter: Kannan Muthukkaruppan
>            Priority: Critical
>              Labels: noob
>         Attachments: metric-stringbuilder-fix.patch
>
>
> I was running some single threaded scan performance tests for a table with small sized rows that is fully cached. Some observations...
> We seem to be doing several wasteful iterations over and/or building of temporary lists.
> 1) One such is the following code in HRegionServer.next():
> {code}
>    boolean moreRows = s.next(values, HRegion.METRIC_NEXTSIZE);
>    if (!values.isEmpty()) {
>      for (KeyValue kv : values) {              ------> #### wasteful in most cases
>        currentScanResultSize += kv.heapSize();
>    }
>    results.add(new Result(values));
> {code}
> By default the "maxScannerResultSize" is Long.MAX_VALUE. In those cases,
> we can avoid the unnecessary iteration to compute currentScanResultSize.
> 2) An example of a wasteful temporary array, is "results" in
> RegionScanner.next().
> {code}
>       results.clear();
>       boolean returnResult = nextInternal(limit, metric);
>       outResults.addAll(results);
> {code}
> results then gets copied over to outResults via an addAll(). Not sure why we can not directly collect the results in outResults.
> 3) Another almost similar exmaple of a wasteful array is "results" in StoreScanner.next(), which eventually also copies its results into "outResults".
> 4) Reduce overhead of "size metric" maintained in StoreScanner.next().
> {code}
>   if (metric != null) {
>      HRegion.incrNumericMetric(this.metricNamePrefix + metric,
>                                copyKv.getLength());
>   }
>   results.add(copyKv);
> {code}
> A single call to next() might fetch a lot of KVs. We can first add up the size of those KVs in a local variable and then in a finally clause increment the metric one shot, rather than updating AtomicLongs for each KV.
> 5) RegionScanner.next() calls a helper RegionScanner.next() on the same object. Both are synchronized methods. Synchronized methods calling nested synchronized methods on the same object are probably adding some small overhead. The inner next() calls isFilterDone() which is a also a synchronized method. We should factor the code to avoid these nested synchronized methods.

--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators: https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira