Messages in this thread | | | Date | Fri, 9 Sep 2011 21:41:28 +0400 | Subject | Re: [PATCH 1/1] CFQ: fix handling 'deep' cfqq | From | Maxim Patlasov <> |
| |
Shaohua,
> So the key problem here is how to detect if a device is fast. Doing > the detection > in the dispatch stage always can't give us correct result. A fast device really > should be requests can be finished in short time. So I have something attached. > In my environment, a hard disk is detected slow and a ssd is detected fast, but > I haven't run any benchmark so far. How do you think about it?
Thanks for the patch, I'll test it in several h/w configurations soon and let you know about results.
As a first thought that comes to mind, this patch would hardly meet needs of h/w raids. Every particular hdd may be not very fast, but being assembled in RAID-0, it's still very beneficial to submit many requests in a row. That's why I made estimation in the dispatch stage: if hdd (or h/w raid) is regularly able to drain deep queue quite fast, we should claim it as 'fast' and avoid idling if possible.
Thanks, Maxim
| |