You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@beam.apache.org by "Kamil Wasilewski (Jira)" <ji...@apache.org> on 2020/09/16 14:54:00 UTC

[jira] [Commented] (BEAM-10643) Grafina dashboard missing per runner nexmark pages

    [ https://issues.apache.org/jira/browse/BEAM-10643?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17197024#comment-17197024 ] 

Kamil Wasilewski commented on BEAM-10643:
-----------------------------------------

[~apilloud], how about something like this? As you can see in the screenshot below, there are up to three data series for each chart: standard, SQL and ZetaSQL. If there is no data for SQL and ZetaSQL variants in the database, only one data series is displayed. Conversely, when adding SQL/ZetaSQL variants of tests, the results should be automatically displayed without any modifications to the dashboard.

Also, it would be no longer possible to display results for multiple runners. The drop-down list of runners would still be there, but only one runner may be selected.

I'm looking forward to your opinion.

 !Zrzut ekranu 2020-09-16 o 16.12.08.png! 

> Grafina dashboard missing per runner nexmark pages
> --------------------------------------------------
>
>                 Key: BEAM-10643
>                 URL: https://issues.apache.org/jira/browse/BEAM-10643
>             Project: Beam
>          Issue Type: Bug
>          Components: testing-nexmark
>            Reporter: Andrew Pilloud
>            Assignee: Kamil Wasilewski
>            Priority: P2
>         Attachments: Zrzut ekranu 2020-09-16 o 16.12.08.png
>
>
> The perfkit dashboard was just turned down, it had a page comparing different implementations for the same query for each runner (Dataflow, Direct, ect). For example, this would be "Java Batch vs Java Streaming vs SQL vs ZetaSQL on Dataflow".
> The new grafina dashboard breaks down nexmark per implementation (Java, SQL, missing ZetaSQL), and hides streaming which isn't a useful comparison as different runners have different sample sizes and measurement techniques.



--
This message was sent by Atlassian Jira
(v8.3.4#803005)