[jira] [Created] (HBASE-15376) ScanNext metric is size-based while every other per-operation metric is time based

["Enis Soztutar (JIRA)" <ji...@apache.org>]

Enis Soztutar created HBASE-15376:
-------------------------------------

             Summary: 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


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)