[
https://issues.apache.org/jira/browse/HBASE-16213?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15415690#comment-15415690
]
ramkrishna.s.vasudevan commented on HBASE-16213:
------------------------------------------------
bq.Also this may not improve E2E throughout or latency much, but may cut down
the CPU usage.
I am not sure I get this. For random reads it is the seek which takes time, if
that is improved then we should see some perf gain in terms of throughput or
latency.
Or you mean the same % will not be felt in E2E performance rather it would be
much more reduced but still visible?
> A new HFileBlock structure for fast random get
> ----------------------------------------------
>
> Key: HBASE-16213
> URL: https://issues.apache.org/jira/browse/HBASE-16213
> Project: HBase
> Issue Type: New Feature
> Components: Performance
> Reporter: binlijin
> Assignee: binlijin
> Attachments: HBASE-16213-master_v1.patch, HBASE-16213.patch,
> HBASE-16213_branch1_v3.patch, HBASE-16213_v2.patch, hfile-cpu.png,
> hfile_block_performance.pptx, hfile_block_performance2.pptx,
> new-hfile-block.xlsx
>
>
> HFileBlock store cells sequential, current when to get a row from the block,
> it scan from the first cell until the row's cell.
> The new structure store every row's start offset with data, so it can find
> the exact row with binarySearch.
> I use EncodedSeekPerformanceTest test the performance.
> First use ycsb write 100w data, every row have only one qualifier, and
> valueLength=16B/64/256B/1k.
> Then use EncodedSeekPerformanceTest to test random read 1w or 100w row, and
> also record HFileBlock's dataSize/dataWithMetaSize in the encoding.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
