[jira] [Commented] (AMBARI-21214) Use a uuid vs long row key for metrics in AMS schema

["Aravindan Vijayan (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/AMBARI-21214?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16054523#comment-16054523 ] 

Aravindan Vijayan commented on AMBARI-21214:
--------------------------------------------

Pushed to branch-3.0-ams

> Use a uuid vs long row key for metrics in AMS schema
> ----------------------------------------------------
>
>                 Key: AMBARI-21214
>                 URL: https://issues.apache.org/jira/browse/AMBARI-21214
>             Project: Ambari
>          Issue Type: Bug
>          Components: ambari-metrics
>    Affects Versions: 3.0.0
>            Reporter: Aravindan Vijayan
>            Assignee: Aravindan Vijayan
>             Fix For: 3.0.0
>
>         Attachments: AMBARI-21214.patch
>
>
> Presently we store (metircName, hostname, serverTime, appId, instanceId) for every row.
> Although this makes SQL query easy to write, it adds duplication when storing it in the Regions memory.
> We should have a unique integer ID for an application-instance metric on a host and use that as a lookup table with a write through cache when inserting or reading data in the METRIC_RECORD table.



--
This message was sent by Atlassian JIRA
(v6.4.14#64029)