[
https://issues.apache.org/jira/browse/MESOS-6630?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16712121#comment-16712121
]
Meng Zhu commented on MESOS-6630:
---------------------------------
{noformat}
commit 4e512618d508f32170193c49d72011189f6e2fa1
Author: Meng Zhu <[email protected]>
Date: Fri Oct 12 17:28:39 2018 -0700
Added an allocator benchmark for quota performance.
This benchmark evaluates the allocator performance in
the presence of roles with both small quota (which can
be satisfied by half an agent) as well as large quota
(which need resources from two agents). We setup the cluster,
trigger one allocation cycle and measure the elapsed time.
Review: https://reviews.apache.org/r/69097
{noformat}
{noformat}
commit 740fa11a33df8528742b3e784206d00111edc4a3
Author: Meng Zhu [email protected]
Date: Fri Oct 19 22:44:04 2018 -0700
Added a benchmark to compare quota and nonquota allocation performance.
This benchmark evaluates the performance difference between nonquota
and quota settings. In both settings, the same allocations are made
for fair comparison. In particular, since the agent will always be
allocated as a whole in nonquota settings, we should also avoid
agent chopping in quota setting as well. Thus in this benchmark,
quotas are only set to be multiples of whole agent resources.
This is also why we have this dedicated benchmark for comparison
rather than extending the existing quota benchmarks (which involves
agent chopping).
Review: https://reviews.apache.org/r/69098
{noformat}
> Add some benchmark test for quota allocation
> --------------------------------------------
>
> Key: MESOS-6630
> URL: https://issues.apache.org/jira/browse/MESOS-6630
> Project: Mesos
> Issue Type: Task
> Components: allocation
> Reporter: Guangya Liu
> Assignee: Meng Zhu
> Priority: Major
> Labels: mesosphere, performance
> Attachments: mesos-master_nonquota_1206.stacks,
> mesos-master_quota_1206.stacks
>
>
> Comparing to non-quota allocation, current quota allocation involves a
> separate allocation stage and additional tracking such as headroom and role
> consumed quota. Thus quota allocation performance could be drastically
> different (probably slower) than non-quota allocation. A dedicated benchmark
> for quota allocation is necessary.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
