[
https://issues.apache.org/jira/browse/HBASE-9910?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14290584#comment-14290584
]
Vikas Vishwakarma commented on HBASE-9910:
------------------------------------------
Hi JM,
I gave it a try and submitting the initial cut for your feedback. Please review
the submitted patch file and let me know your thoughts on this.
I have added support to specify codec and cipher for HFile writer/reader tests
and also included test combinations for gz codec and aes cipher similar to what
is being done in TestHFilePerformance. For each combination writer will run
once followed by all the readers.
While looking at TestHFilePerformance, I realized that AES cipher tests were
completely broken (HBASE-12866) so I have fixed that while including it in
HFilePerformanceEvaluation
Also I have added test summary and I am deleting all the test files at the end
of the test.
I have tested it locally using bin/start-hbase.sh;bin/hbase
org.apache.hadoop.hbase.HFilePerformanceEvaluation
Sample Test Result:
========================
***************
Result Summary
***************
Running SequentialWriteBenchmark with codec[none] cipher[none] for 1000000 rows
took 761ms.
Running UniformRandomSmallScan with codec[none] cipher[none] for 1000000 rows
took 2271ms.
Running UniformRandomReadBenchmark with codec[none] cipher[none] for 1000000
rows took 71717ms.
Running GaussianRandomReadBenchmark with codec[none] cipher[none] for 1000000
rows took 79260ms.
Running SequentialReadBenchmark with codec[none] cipher[none] for 1000000 rows
took 227ms.
Running SequentialWriteBenchmark with codec[gz] cipher[none] for 1000000 rows
took 1025ms.
Running UniformRandomSmallScan with codec[gz] cipher[none] for 1000000 rows
took 15829ms.
Running UniformRandomReadBenchmark with codec[gz] cipher[none] for 1000000 rows
took 145314ms.
Running GaussianRandomReadBenchmark with codec[gz] cipher[none] for 1000000
rows took 155687ms.
Running SequentialReadBenchmark with codec[gz] cipher[none] for 1000000 rows
took 434ms.
Running SequentialWriteBenchmark with codec[none] cipher[aes] for 1000000 rows
took 953ms.
Running UniformRandomSmallScan with codec[none] cipher[aes] for 1000000 rows
took 7113ms.
Running UniformRandomReadBenchmark with codec[none] cipher[aes] for 1000000
rows took 121273ms.
Running GaussianRandomReadBenchmark with codec[none] cipher[aes] for 1000000
rows took 134818ms.
Running SequentialReadBenchmark with codec[none] cipher[aes] for 1000000 rows
took 378ms.
Running SequentialWriteBenchmark with codec[gz] cipher[aes] for 1000000 rows
took 1187ms.
Running UniformRandomSmallScan with codec[gz] cipher[aes] for 1000000 rows took
15546ms.
Running UniformRandomReadBenchmark with codec[gz] cipher[aes] for 1000000 rows
took 158620ms.
Running GaussianRandomReadBenchmark with codec[gz] cipher[aes] for 1000000 rows
took 176853ms.
Running SequentialReadBenchmark with codec[gz] cipher[aes] for 1000000 rows
took 506ms.
> TestHFilePerformance and HFilePerformanceEvaluation should be merged in a
> single HFile performance test class.
> --------------------------------------------------------------------------------------------------------------
>
> Key: HBASE-9910
> URL: https://issues.apache.org/jira/browse/HBASE-9910
> Project: HBase
> Issue Type: Bug
> Components: Performance, test
> Reporter: Jean-Marc Spaggiari
> Assignee: Vikas Vishwakarma
> Attachments: HBASE-9910.patch
>
>
> Today TestHFilePerformance and HFilePerformanceEvaluation are doing slightly
> different kind of performance tests both for the HFile. We should consider
> merging those 2 tests in a single class.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
