[
https://issues.apache.org/jira/browse/HBASE-15376?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15176780#comment-15176780
]
Enis Soztutar commented on HBASE-15376:
---------------------------------------
I was checking whether we can change the metric names and/or remove metrics in
minor or major versions. The
https://hbase.apache.org/book.html#hbase.versioning page says that there is no
operational compatibility between minor versions, so according to the
previously agreed on guidelines, we can (but do not need to) in fact change the
name of the metrics in 1.3+.
Coming to this patch, I think we should rename the metric name "ScanNext" to
"ScanSize" and add "ScanTime" as the new metric. We should do this for the
per-region metric as well. cc [~eclark], [~saint....@gmail.com].
> ScanNext metric is size-based while every other per-operation metric is time
> based
> ----------------------------------------------------------------------------------
>
> Key: HBASE-15376
> URL: https://issues.apache.org/jira/browse/HBASE-15376
> Project: HBase
> Issue Type: Sub-task
> Reporter: Enis Soztutar
> Attachments: HBASE-15376.patch
>
>
> We have per-operation metrics for {{Get}}, {{Mutate}}, {{Delete}},
> {{Increment}}, and {{ScanNext}}.
> The metrics are emitted like:
> {code}
> "Get_num_ops" : 4837505,
> "Get_min" : 0,
> "Get_max" : 296,
> "Get_mean" : 0.2934618155433431,
> "Get_median" : 0.0,
> "Get_75th_percentile" : 0.0,
> "Get_95th_percentile" : 1.0,
> "Get_99th_percentile" : 1.0,
> ...
> "ScanNext_num_ops" : 194705,
> "ScanNext_min" : 0,
> "ScanNext_max" : 18441,
> "ScanNext_mean" : 7468.274651395701,
> "ScanNext_median" : 583.0,
> "ScanNext_75th_percentile" : 583.0,
> "ScanNext_95th_percentile" : 13481.0,
> "ScanNext_99th_percentile" : 13481.0,
> {code}
> The problem is that all of Get,Mutate,Delete,Increment,Append,Replay are time
> based tracking how long the operation ran, while ScanNext is tracking
> returned response sizes (returned cell-sizes to be exact). Obviously, this is
> very confusing and you would only know this subtlety if you read the metrics
> collection code.
> Not sure how useful is the ScanNext metric as it is today. We can deprecate
> it, and introduce a time based one to keep track of scan request latencies.
> ps. Shamelessly using the parent jira (since these seem relavant).
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
