[tarantool-patches] Re: [PATCH v2 09/11] vinyl: do not account zero dump bandwidth

Konstantin Osipov kostja at tarantool.org
Tue Oct 23 10:02:49 MSK 2018

* Vladimir Davydov <vdavydov.dev at gmail.com> [18/10/17 12:14]:
> On Tue, Oct 16, 2018 at 09:25:18PM +0300, Konstantin Osipov wrote:
> > * Vladimir Davydov <vdavydov.dev at gmail.com> [18/10/12 21:23]:
> > > Turns out this isn't enough. We'd better not account too small dumps,
> > > because such dumps have too high overhead associated with file creation.
> > > Our tests create a lot of small dumps using box.snapshot(). Taking them
> > > into account may slow down the overall test execution time or even break
> > > some tests. Let's ignore all dumps of size less than 1 MB for bandwidth
> > > estimation. The new patch is below.
> > 
> > Let's talk face to face, but this entire clutch seems to be
> > becoming too huge with this patch.
> What's wrong with it? If you create a 100 MB file then the time it takes
> to allocate an inode is negligible comparing to the write time. This is
> a typical use case. However, if you create a 1 KB file, then you'll
> spend much more time on inode bookkeeping than actually writing
> anything. This is unlikely in practice (who'd write such small files?),
> but this is done very often by our tests. This results in enabling
> throttling and slowing down test execution time.

Simply put, we need to disable throttling because things get worse
when it's enabled. 

And we simply disable throttling considering this situation an
edge case rather than try to understand what's wrong with our
model that it makes things worse on edges.

Konstantin Osipov, Moscow, Russia, +7 903 626 22 32
http://tarantool.io - www.twitter.com/kostja_osipov

More information about the Tarantool-patches mailing list