Sangjin Lee commented on YARN-2556:

[~lichangleo], thanks much for updating the patch! I'm sorry I didn't get to 
review this sooner. I'll spend some time today to review this in more detail.

That said, I'd like to raise one important point as a goal: *it would be great 
if we can have a single TimelineServicePerformance class that can accommodate 
both v1 and v2 timeline service tests*. That way, any v1- or v2-specific code 
can be isolated to the specific writers. If we came this close but are still 
left with two versions of TimelineServicePerformance, that would not be such a 
great outcome.

I think the latest patch is almost there. We might need to make the writer base 
class more generic, as it still involves types that are specific to v1. Perhaps 
we can simply use the mapper class in TimelineServicePerformance (i.e. remove 
EntityWriter altogether from TimelineServicePerformance), and move any v1- or 
v2-specific writers into their own classes outside TimelineServicePerformance.

I know the option handling is going to be a sticky point, as it will be 
difficult to have specific mappers inject their own option handling. Maybe 
that's something we might have to live with. I'm curious if you have ideas to 
make even that generic.

> Tool to measure the performance of the timeline server
> ------------------------------------------------------
>                 Key: YARN-2556
>                 URL: https://issues.apache.org/jira/browse/YARN-2556
>             Project: Hadoop YARN
>          Issue Type: Sub-task
>          Components: timelineserver
>            Reporter: Jonathan Eagles
>            Assignee: Chang Li
>              Labels: BB2015-05-TBR
>         Attachments: YARN-2556-WIP.patch, YARN-2556-WIP.patch, 
> YARN-2556.1.patch, YARN-2556.10.patch, YARN-2556.2.patch, YARN-2556.3.patch, 
> YARN-2556.4.patch, YARN-2556.5.patch, YARN-2556.6.patch, YARN-2556.7.patch, 
> YARN-2556.8.patch, YARN-2556.9.patch, YARN-2556.patch, yarn2556.patch, 
> yarn2556.patch, yarn2556_wip.patch
> We need to be able to understand the capacity model for the timeline server 
> to give users the tools they need to deploy a timeline server with the 
> correct capacity.
> I propose we create a mapreduce job that can measure timeline server write 
> and read performance. Transactions per second, I/O for both read and write 
> would be a good start.
> This could be done as an example or test job that could be tied into gridmix.

This message was sent by Atlassian JIRA

Reply via email to