mesos-reviews mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Benjamin Mahler <bmah...@apache.org>
Subject Re: Review Request 68731: Added a ScalarResourceQuantities type to improve sorters performance.
Date Fri, 21 Sep 2018 23:03:31 GMT


> On Sept. 19, 2018, 9:27 p.m., Meng Zhu wrote:
> > src/master/allocator/sorter/sorter.hpp
> > Lines 163-164 (patched)
> > <https://reviews.apache.org/r/68731/diff/2/?file=2090164#file2090164line163>
> >
> >     Why we are enforcing alphabetically sorted? Seems to me unnecessary, especially
we do not force it.

We don't rely on it yet, but it will be beneficial for arithmetic operations. I'll add a comment
for that.


> On Sept. 19, 2018, 9:27 p.m., Meng Zhu wrote:
> > src/master/allocator/sorter/sorter.hpp
> > Lines 194 (patched)
> > <https://reviews.apache.org/r/68731/diff/2/?file=2090164#file2090164line194>
> >
> >     I think it is a good idea to print the whole vector as well.

I added a TODO, it's a little verbose unless we have a `stringify(const pair<T1, T2>&
p)` overload.


- Benjamin


-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/68731/#review208774
-----------------------------------------------------------


On Sept. 18, 2018, 9:20 p.m., Benjamin Mahler wrote:
> 
> -----------------------------------------------------------
> This is an automatically generated e-mail. To reply, visit:
> https://reviews.apache.org/r/68731/
> -----------------------------------------------------------
> 
> (Updated Sept. 18, 2018, 9:20 p.m.)
> 
> 
> Review request for mesos, Gastón Kleiman and Meng Zhu.
> 
> 
> Bugs: MESOS-9239
>     https://issues.apache.org/jira/browse/MESOS-9239
> 
> 
> Repository: mesos
> 
> 
> Description
> -------
> 
> This type replaces the use of hashmaps keyed by resource names in
> favor of storing vectors of `pair<string,Value::Scalar>`, in order
> to avoid the performance penalty of using hashmaps.
> 
> Running *HierarchicalAllocator_BENCHMARK_Test.DeclineOffers/21 shows
> the following improvement:
> 
> Using 10000 agents and 1000 frameworks
> Added 1000 frameworks in 42.49ms -> 42.85ms (no change)
> Added 10000 agents in 7.69secs -> 4.89secs (normalized: 1 -> 0.64)
> round 0 allocate() took 5.42secs -> 3.53secs (nomralized: 1 -> 0.65)
> 
> 
> Diffs
> -----
> 
>   src/master/allocator/sorter/drf/sorter.hpp 5a4fa5e2dca61168923261230b1f5c245354cbb7

>   src/master/allocator/sorter/random/sorter.hpp 7f6c0de70e3ae03d7362fb9e140b93435e530499

>   src/master/allocator/sorter/sorter.hpp 432ccfe24ed2854df9cc186a8691009cbdb763c7 
> 
> 
> Diff: https://reviews.apache.org/r/68731/diff/2/
> 
> 
> Testing
> -------
> 
> make check
> 
> 
> Thanks,
> 
> Benjamin Mahler
> 
>


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