hbase-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Amit Patel (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (HBASE-18542) [HLC] Performance microbenchmarks
Date Tue, 22 Aug 2017 03:29:00 GMT

    [ https://issues.apache.org/jira/browse/HBASE-18542?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16136211#comment-16136211

Amit Patel commented on HBASE-18542:

Added patch [^HBASE-18542.HBASE-14070.HLC.001.patch] that benchmarks performance of Clock#now
and Clock#update under both single threaded and multi threaded cases. Below is a summary of
the performance (where the number is the multiplicative factor of how much longer it took
compared to the system clock). 

||System clock||System monotonic|Hybrid logical clock||
|Clock#now single threaded|1.01|1.41|
|Clock#update single threaded|1.01|1.36|
|Clock#now multi threaded|1.23|1.76|
|Clock#update multi threaded|2.28|2.20|

> [HLC] Performance microbenchmarks
> ---------------------------------
>                 Key: HBASE-18542
>                 URL: https://issues.apache.org/jira/browse/HBASE-18542
>             Project: HBase
>          Issue Type: Sub-task
>            Reporter: Appy
>            Assignee: Amit Patel
>         Attachments: HBASE-18542.HBASE-14070.HLC.001.patch
> Need tests to benchmark performance of Clock#now() and update() functions (for all types
of clocks).
> If update() is too costly, we can do optimizations in ExecuteProceduresRemoteCall#call(),
HRegion#replayRecoveredEdits() and other places where we call update() in loop. Instead, it
might be faster to calculate max timestamp in loop and call update() just once.

This message was sent by Atlassian JIRA

View raw message