Re: RFC on io-stalls patch

[Marcelo Tosatti]

On Mon, 14 Jul 2003, Jens Axboe wrote:

> On Sun, Jul 13 2003, Andrea Arcangeli wrote:
> > On Sun, Jul 13, 2003 at 11:01:16AM +0200, Jens Axboe wrote:
> > > No I don't have anything specific, it just seems like a bad heuristic to
> > > get rid of. I can try and do some testing tomorrow. I do feel strongly
> >
> > well, it's not an heuristic, it's a simplification and it will certainly
> > won't provide any benefit (besides saving some hundred kbytes of ram per
> > harddisk that is a minor benefit).
>
> You are missing my point - I don't care about loosing the extra request
> list, I never said anything about that in this thread. I care about
> loosing the reserved requests for reads. And we can do that just fine
> with just holding back a handful of requests.
>
> > > that we should at least make sure to reserve a few requests for reads
> > > exclusively, even if you don't agree with the oversized check. Anything
> > > else really contradicts all the io testing we have done the past years
> > > that shows how important it is to get a read in ASAP. And doing that in
> >
> > Important for latency or throughput? Do you know which is the benchmarks
> > that returned better results with the two queues, what's the theory
> > behind this?
>
> Forget the two queues, noone has said anything about that. The reserved
> reads are important for latency reasons, not throughput.

So Jens,

Please bench (as you said you would), and send us the results.

Its very important.

Thanks