mesos-reviews mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Meng Zhu <m...@mesosphere.io>
Subject Re: Review Request 69097: Added an allocator benchmark for quota performance.
Date Thu, 06 Dec 2018 01:16:10 GMT


> On Nov. 2, 2018, 3:22 p.m., Benjamin Mahler wrote:
> > src/tests/hierarchical_allocator_benchmarks.cpp
> > Lines 498-499 (patched)
> > <https://reviews.apache.org/r/69097/diff/1/?file=2101620#file2101620line498>
> >
> >     Why not make it even smaller (e.g. 1/4 or 1/10th) to be more pathological?

Updated it to 1/5 for small roles and 5 for large roles.


> On Nov. 2, 2018, 3:22 p.m., Benjamin Mahler wrote:
> > src/tests/hierarchical_allocator_benchmarks.cpp
> > Lines 601-609 (patched)
> > <https://reviews.apache.org/r/69097/diff/1/?file=2101620#file2101620line601>
> >
> >     No need for the comments, seems self explanatory?

Done.


- Meng


-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/69097/#review210291
-----------------------------------------------------------


On Oct. 22, 2018, 3:32 p.m., Meng Zhu wrote:
> 
> -----------------------------------------------------------
> This is an automatically generated e-mail. To reply, visit:
> https://reviews.apache.org/r/69097/
> -----------------------------------------------------------
> 
> (Updated Oct. 22, 2018, 3:32 p.m.)
> 
> 
> Review request for mesos, Benjamin Mahler and Gastón Kleiman.
> 
> 
> Bugs: MESOS-6630
>     https://issues.apache.org/jira/browse/MESOS-6630
> 
> 
> Repository: mesos
> 
> 
> Description
> -------
> 
> 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.
> 
> 
> Diffs
> -----
> 
>   src/tests/hierarchical_allocator_benchmarks.cpp bf9167b63747f7b8a402d950947028436307082a

> 
> 
> Diff: https://reviews.apache.org/r/69097/diff/1/
> 
> 
> Testing
> -------
> 
> Result from optimized build on a multicore 2.2GHz machine:
> 
> Benchmark setup: 20 agents, 25 roles, 25 frameworks
> Made 30 allocations in 11.01225ms
> Made 0 allocation in 5.758192ms
> 
> Benchmark setup: 200 agents, 250 roles, 250 frameworks
> Made 300 allocations in 337.974387ms
> Made 0 allocation in 313.298636ms
> 
> Benchmark setup: 2000 agents, 2500 roles, 2500 frameworks
> Made 3000 allocations in 27.881684393secs
> Made 0 allocation in 25.013398926secs
> 
> 
> Thanks,
> 
> Meng Zhu
> 
>


Mime
  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message