Whitepaper: database workloads using RBD with dm-cache and Ceph cache tiering

Whitepaper: database workloads using RBD with dm-cache and Ceph cache tiering

[Douglas Fuller]

I’ve published a whitepaper detailing configuration and benchmark results using SysBench MySQL on top of RBD with dm-cache. Results using a Ceph cache tier are also included. Questions, comments, criticism, nitpicking, and/or discussion are welcome.

http://tracker.ceph.com/projects/ceph/wiki/Ceph_and_dm-cache_for_Database_Workloads

—Doug--
To unsubscribe from this list: send the line "unsubscribe ceph-devel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: Whitepaper: database workloads using RBD with dm-cache and Ceph cache tiering

[Blair Bethwaite]

Hi Douglas,

It's not clear whether you used SSD journals for the OSDs in the base
config (or in the backing pool for the tiered tests) - it looks like
not?

If that's the case then I don't think this provides the most
interesting point of comparison, because the tests are write-latency
sensitive you'd naturally expect SSD journals to make a lot of
difference and would surely start with those in the base config. I'm
interested in how a similar setup would perform compared to regular
journaling and Ceph tiering all on the server-side, e.g., dm-cache (or
bcache) based OSDs - apparently a bunch of people do this in the wild
with success, but AFAIK there has never been any detailed benchmarking
published.

Cheers,

On 3 November 2015 at 05:09, Douglas Fuller <dfuller@redhat.com> wrote:
> I’ve published a whitepaper detailing configuration and benchmark results using SysBench MySQL on top of RBD with dm-cache. Results using a Ceph cache tier are also included. Questions, comments, criticism, nitpicking, and/or discussion are welcome.
>
> http://tracker.ceph.com/projects/ceph/wiki/Ceph_and_dm-cache_for_Database_Workloads
>
> —Doug--
> To unsubscribe from this list: send the line "unsubscribe ceph-devel" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html



-- 
Cheers,
~Blairo
--
To unsubscribe from this list: send the line "unsubscribe ceph-devel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: Whitepaper: database workloads using RBD with dm-cache and Ceph cache tiering

[Douglas Fuller]

Hi Blario,

Thanks for the feedback.

> On Nov 3, 2015, at 9:27 PM, Blair Bethwaite <blair.bethwaite@gmail.com> wrote:
> 
> Hi Douglas,
> 
> It's not clear whether you used SSD journals for the OSDs in the base
> config (or in the backing pool for the tiered tests) - it looks like
> not?

Correct, these tests used spinning disk journals; the test cluster isn’t equipped with SSDs. It may be interesting to know how using NVMe devices for journaling as opposed to a cache tier performs. I’ll run the numbers.

> If that's the case then I don't think this provides the most
> interesting point of comparison, because the tests are write-latency
> sensitive you'd naturally expect SSD journals to make a lot of
> difference and would surely start with those in the base config. I'm
> interested in how a similar setup would perform compared to regular
> journaling and Ceph tiering all on the server-side, e.g., dm-cache (or
> bcache) based OSDs - apparently a bunch of people do this in the wild
> with success, but AFAIK there has never been any detailed benchmarking
> published.

I haven’t heard of this type of configuration myself, but it might be a possible scenario to explore.

Thanks again,
—Doug

> 
> Cheers,
> 
> On 3 November 2015 at 05:09, Douglas Fuller <dfuller@redhat.com> wrote:
>> I’ve published a whitepaper detailing configuration and benchmark results using SysBench MySQL on top of RBD with dm-cache. Results using a Ceph cache tier are also included. Questions, comments, criticism, nitpicking, and/or discussion are welcome.
>> 
>> http://tracker.ceph.com/projects/ceph/wiki/Ceph_and_dm-cache_for_Database_Workloads
>> 
>> —Doug--
>> To unsubscribe from this list: send the line "unsubscribe ceph-devel" in
>> the body of a message to majordomo@vger.kernel.org
>> More majordomo info at  http://vger.kernel.org/majordomo-info.html
> 
> 
> 
> -- 
> Cheers,
> ~Blairo
> --
> To unsubscribe from this list: send the line "unsubscribe ceph-devel" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html

--
To unsubscribe from this list: send the line "unsubscribe ceph-devel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html