[
https://issues.apache.org/jira/browse/HBASE-13291?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14382801#comment-14382801
]
Andrew Purtell commented on HBASE-13291:
----------------------------------------
bq. it fingers the HBB#_get again.
HBB is generated from something called Heap-X-Buffer.java.template when the JDK
is built. However, looks like _get for HBB is:
{code}
byte _get(int i) { // package-private
return hb[i];
}
{code}
... a simple array access. You'd expect this along with a lot of comparator
work if we're scanning or skipping through a lot of cells in memory. I think
the way to optimize this is keep an index to cells in the blockcache, to
eliminate the need to read a length, skip, read a length, skip, read a
length...
> Lift the scan ceiling
> ---------------------
>
> Key: HBASE-13291
> URL: https://issues.apache.org/jira/browse/HBASE-13291
> Project: HBase
> Issue Type: Improvement
> Components: Scanners
> Affects Versions: 1.0.0
> Reporter: stack
> Assignee: stack
> Attachments: 13291.inlining.txt, Screen Shot 2015-03-26 at 12.12.13
> PM.png, q (1).png, traces.7.svg, traces.filterall.svg, traces.nofilter.svg,
> traces.small2.svg, traces.smaller.svg
>
>
> Scanning medium sized rows with multiple concurrent scanners exhibits
> interesting 'ceiling' properties. A server runs at about 6.7k ops a second
> using 450% of possible 1600% of CPUs when 4 clients each with 10 threads
> doing scan 1000 rows. If I add '--filterAll' argument (do not return
> results), then we run at 1450% of possible 1600% possible but we do 8k ops a
> second.
> Let me attach flame graphs for two cases. Unfortunately, there is some
> frustrating dark art going on. Let me try figure it... Filing issue in
> meantime to keep score in.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
