any plan for sbd upstream?

[min...@gmail.com]

Hi list,

Is there a plan to push sbd code to upstream Linux kernel?

Regards,

Ming

[Liu Yuan]

On Fri, Nov 11, 2016 at 11:34:32PM -0800, min...@gmail.com wrote:
> Hi list,
>
> Is there a plan to push sbd code to upstream Linux kernel?

No. why bother upstreaming it? From Userland, we enjoy the fast modification and better portability.

Yuan

[min...@gmail.com]

I tried to test sheepdog's performance with super fast NVMe drive.

"sbd"(similar as ceph's rbd) should be the best way to test the performance.

Right?

 

Yuan

[Liu Yuan]

On Wed, Nov 16, 2016 at 10:32:22AM -0800, min...@gmail.com wrote:
> On Sunday, November 13, 2016 at 11:45:53 PM UTC-8, Liu Yuan wrote:
> >

> > On Fri, Nov 11, 2016 at 11:34:32PM -0800, min...@gmail.com <javascript:>

> > wrote:
> > > Hi list,
> > >
> > > Is there a plan to push sbd code to upstream Linux kernel?
> >
> > No. why bother upstreaming it? From Userland, we enjoy the fast
> > modification and better portability.
> >
>
> I tried to test sheepdog's performance with super fast NVMe drive.
> "sbd"(similar as ceph's rbd) should be the best way to test the performance.
>
> Right?

Yes in theory. But practically as someone reported that qemu driver perf is on a
par with the sbd. I guess sbd need some enhancements (make use of blk-mq, etc.)
to get the best of the performance.

I think there is some work to be done to make best out of NVMe drive.

Yuan