aurora-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "David McLaughlin (JIRA)" <j...@apache.org>
Subject [jira] [Updated] (AURORA-471) Measure performance of the getTasks API call
Date Fri, 23 May 2014 20:29:01 GMT

     [ https://issues.apache.org/jira/browse/AURORA-471?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]

David McLaughlin updated AURORA-471:
------------------------------------

    Attachment: jobsummary-profile.png
                browser-profile.png

Attaching a browser profile which confirms the browser is mostly waiting for the server to
respond. 

I was going to try to do in-depth timing on the server for getTasks, but then I noticed that
getJobSummary itself calls getTasks (and other API calls) and this executes in only 850ms.
The major difference is the size of the uncompressed response - 10MB versus 1.7MB. This would
infer that the vast majority of the time on the server is spent in thrift serialization. 


> Measure performance of the getTasks API call
> --------------------------------------------
>
>                 Key: AURORA-471
>                 URL: https://issues.apache.org/jira/browse/AURORA-471
>             Project: Aurora
>          Issue Type: Sub-task
>          Components: Scheduler, UI
>            Reporter: David McLaughlin
>            Assignee: David McLaughlin
>         Attachments: browser-profile.png, jobsummary-profile.png
>
>
> We are trying to improve the performance of the job page in the new UI. Preliminary profiling
shows that most of the time is spent on the thrift.js call to getTasks. Before suggesting
API improvements, we need to determine where time is being spent in the request. 



--
This message was sent by Atlassian JIRA
(v6.2#6252)

Mime
View raw message