Re: Improving connection scalability (src/backend/storage/ipc/procarray.c)

Em sex., 27 de mai. de 2022 às 18:08, Andres Freund <andres(at)anarazel(dot)de>
escreveu:

> Hi,
>
> On 2022-05-27 03:30:46 +0200, Tomas Vondra wrote:
> > On 5/27/22 02:11, Ranier Vilela wrote:
> > > ./pgbench -M prepared -c $conns -j $conns -T 60 -S -n -U postgres
> > >
> > > pgbench (15beta1)
> > > transaction type: <builtin: select only>
> > > scaling factor: 1
> > > query mode: prepared
> > > number of clients: 100
> > > number of threads: 100
> > > maximum number of tries: 1
> > > duration: 60 s
> > >
> > > conns tps head tps patched
> > >
> > > 1 17126.326108 17792.414234
> > > 10 82068.123383 82468.334836
> > > 50 73808.731404 74678.839428
> > > 80 73290.191713 73116.553986
> > > 90 67558.483043 68384.906949
> > > 100 65960.982801 66997.793777
> > > 200 62216.011998 62870.243385
> > > 300 62924.225658 62796.157548
> > > 400 62278.099704 63129.555135
> > > 500 63257.930870 62188.825044
> > > 600 61479.890611 61517.913967
> > > 700 61139.354053 61327.898847
> > > 800 60833.663791 61517.913967
> > > 900 61305.129642 61248.336593
> > > 1000 60990.918719 61041.670996
> > >
> >
> > These results look much saner, but IMHO it also does not show any clear
> > benefit of the patch. Or are you still claiming there is a benefit?
>
> They don't look all that sane to me - isn't that way lower than one would
> expect?

Yes, quite disappointing.

Restricting both client and server to the same four cores, a
> thermically challenged older laptop I have around I get 150k tps at both 10
> and 100 clients.
>
And you can share the benchmark details? Hardware, postgres and pgbench,
please?

>
> Either way, I'd not expect to see any GetSnapshotData() scalability
> effects to
> show up on an "Intel® Core™ i5-8250U CPU Quad Core" - there's just not
> enough
> concurrency.
>
This means that our customers will not see any connections scalability with
PG15, using the simplest hardware?

> The correct pieces of these changes seem very unlikely to affect
> GetSnapshotData() performance meaningfully.
>
> To improve something like GetSnapshotData() you first have to come up with
> a
> workload that shows it being a meaningful part of a profile. Unless it is,
> performance differences are going to just be due to various forms of noise.
>
Actually in the profiles I got with perf, GetSnapShotData() didn't show up.

regards,
Ranier Vilela