hbase-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "jiraposter@reviews.apache.org (Commented) (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (HBASE-4516) HFile-level load tester with compaction and random-read workloads
Date Fri, 30 Sep 2011 00:07:46 GMT

    [ https://issues.apache.org/jira/browse/HBASE-4516?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13117767#comment-13117767
] 

jiraposter@reviews.apache.org commented on HBASE-4516:
------------------------------------------------------


-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/2122/#review2195
-----------------------------------------------------------

Ship it!


Answer my question above M and I'll exclude on commit.  Will add a little note to the 'book'
too that this tool now exists.  Nice.

- Michael


On 2011-09-29 23:54:05, Mikhail Bautin wrote:
bq.  
bq.  -----------------------------------------------------------
bq.  This is an automatically generated e-mail. To reply, visit:
bq.  https://reviews.apache.org/r/2122/
bq.  -----------------------------------------------------------
bq.  
bq.  (Updated 2011-09-29 23:54:05)
bq.  
bq.  
bq.  Review request for hbase, Michael Stack and Jonathan Gray.
bq.  
bq.  
bq.  Summary
bq.  -------
bq.  
bq.  This is a load testing tool for HFile implementations, which supports two workloads:
bq.  
bq.      Compactions (merge the input HFiles). A special case of this is only one input, which
allows to do HFile format conversions.
bq.      Random reads. Launches the specified number of threads that do seeks and short scans
on randomly generated keys.
bq.  
bq.  The original purpose of this tool was to ensure that HFile format v2 did not introduce
performance regressions.
bq.  
bq.  Keys for the read workload are generated randomly between the first and the last key
of the HFile. At each position, instead of precisely calculating the correct probability for
every byte value b, we select a uniformly random byte between in the allowed [low, high] range.
In addition, there is a heuristic that determines the positions at which the key has hex characters,
and the random key contains hex characters at those positions as well.
bq.  
bq.  Example output for the random read workload:
bq.  Time: 120 sec, seek/sec: 8290, kv/sec: 30351, kv bytes/sec: 91868121, blk/sec: 10147,
unique keys: 232779
bq.  
bq.  Also refactoring and clarifying the confusing situation when a StoreFile happens to have
a different Bloom filter type than what is configured for the column family.
bq.  
bq.  
bq.  This addresses bug HBASE-4516.
bq.      https://issues.apache.org/jira/browse/HBASE-4516
bq.  
bq.  
bq.  Diffs
bq.  -----
bq.  
bq.    src/main/java/org/apache/hadoop/hbase/io/hfile/HFileBlock.java b429819 
bq.    src/main/java/org/apache/hadoop/hbase/regionserver/StoreFile.java fd9e7ef 
bq.    src/test/java/org/apache/hadoop/hbase/regionserver/HFileReadWriteTest.java PRE-CREATION

bq.  
bq.  Diff: https://reviews.apache.org/r/2122/diff
bq.  
bq.  
bq.  Testing
bq.  -------
bq.  
bq.  Running unit tests and the load tester tool itself.
bq.  
bq.  
bq.  Thanks,
bq.  
bq.  Mikhail
bq.  
bq.


                
> HFile-level load tester with compaction and random-read workloads
> -----------------------------------------------------------------
>
>                 Key: HBASE-4516
>                 URL: https://issues.apache.org/jira/browse/HBASE-4516
>             Project: HBase
>          Issue Type: Test
>            Reporter: Mikhail Bautin
>            Priority: Minor
>             Fix For: 0.94.0
>
>
> This is a load testing tool for HFile implementations, which supports two workloads:
> - Compactions (merge the input HFiles). A special case of this is only one input, which
allows to do HFile format conversions.
> - Random reads. Launches the specified number of threads that do seeks and short scans
on randomly generated keys.
> The original purpose of this tool was to ensure that HFile format v2 did not introduce
performance regressions.
> Keys for the read workload are generated randomly between the first and the last key
of the HFile. At each position, instead of precisely calculating the correct probability for
every byte value b, we select a uniformly random byte between in the allowed [low, high] range.
In addition, there is a heuristic that determines the positions at which the key has hex characters,
and the random key contains hex characters at those positions as well.
> Example output for the random read workload:
> Time: 120 sec, seek/sec: 8290, kv/sec: 30351, kv bytes/sec: 91868121, blk/sec: 10147,
unique keys: 232779

--
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

        

Mime
View raw message