Messages in this thread | | | Subject | Re: [PATCH] virtio_blk: use noop elevator by default | From | Fernando Luis Vázquez Cao <> | Date | Wed, 27 Aug 2008 14:14:07 +0900 |
| |
On Tue, 2008-08-26 at 16:39 +0200, Jens Axboe wrote: > On Tue, Aug 26 2008, Fernando Luis Vázquez Cao wrote: > > Hi Rusty, > > > > Would it make sense to use noop by default? After all we do not know > > what is behind the backend driver and the hypervisor is likely to do its > > own scheduling anyway. I guess this is the reason the Xen guys took this > > approach. > > > > What do you think about the patch below? > > I plan to include some variant of disk profiling for 2.6.28 which will > let eg CFQ turn off idling for such device types, I think that is a > better solution.
Hi Jens,
That is good news. With the proliferation of intelligent disk controllers and SSDs, the disk profiling approach seems to be the right way to go in general and I think it was badly needed.
>From your example my wild guess is that disk profiling will be a I/O controller-specific auto-tuning feature, not a new functionality of the generic elevator layer. Is this interpretation correct? Would it make sense in some cases to change elevators automatically depending on the characteristics of the underlying device instead (e.g. we might not need any of the extra features CFQ provides, for example)?
I would like to take a look at those patches so I peeked into your git tree, but I could not find them (I probably chose the wrong branches). Are they accessible through your kernel.org's git repository?
Thanks!
Fernando
-- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
| |