Are there significant performance enhancements in 0.56.x to be expected soon or planned in the near future?

Are there significant performance enhancements in 0.56.x to be expected soon or planned in the near future?

[Lachfeld, Jutta]

Hi all,

in expectation of better performance, we are just switching from CEPH version 0.48 to 0.56.1
for comparisons between Hadoop with HDFS and Hadoop with CEPH FS.

We are now wondering whether there are currently any development activities 
concerning further significant performance enhancements, 
or whether further significant performance enhancements are already planned for the near future.

I would now be loath to start benchmarking with 0.56.1 and then, a month or so later, detect that there have been significant performance enhancements in CEPH in the meantime.

Regards,
Jutta.

-
jutta.lachfeld@ts.fujitsu.com, Fujitsu Technology Solutions PBG PDG ES&S SWE SOL 4, "Infrastructure Solutions", MchD 5B, Tel. ..49-89-3222-2705, Company Details: http://de.ts.fujitsu.com/imprint

Re: Are there significant performance enhancements in 0.56.x to be expected soon or planned in the near future?

[Wido den Hollander]

On 01/09/2013 01:51 PM, Lachfeld, Jutta wrote:
> Hi all,
>
> in expectation of better performance, we are just switching from CEPH version 0.48 to 0.56.1
> for comparisons between Hadoop with HDFS and Hadoop with CEPH FS.
>
> We are now wondering whether there are currently any development activities
> concerning further significant performance enhancements,
> or whether further significant performance enhancements are already planned for the near future.
>

Yes, 0.56(.1) has a significant performance increase compared to 0.48

Two blogposts which might be interesting to read:
* http://ceph.com/dev-notes/whats-new-in-the-land-of-osd/
* http://ceph.com/uncategorized/argonaut-vs-bobtail-performance-preview/

I'm not running with HDFS, but I see a good performance increase with 
Virtual Machines running on RBD.

Wido

> I would now be loath to start benchmarking with 0.56.1 and then, a month or so later, detect that there have been significant performance enhancements in CEPH in the meantime.
>
> Regards,
> Jutta.
>
> -
> jutta.lachfeld@ts.fujitsu.com, Fujitsu Technology Solutions PBG PDG ES&S SWE SOL 4, "Infrastructure Solutions", MchD 5B, Tel. ..49-89-3222-2705, Company Details: http://de.ts.fujitsu.com/imprint
> --
> 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: Are there significant performance enhancements in 0.56.x to be expected soon or planned in the near future?

[Christopher Kunz]

Hi,
> 
> Yes, 0.56(.1) has a significant performance increase compared to 0.48
> 
That is not exactly the OP's question, though. If I understand
correctly, she is concerned about ongoing performance improvements
within the "bobtail" branch, i.e. between 0.56.1 and 0.56.X (with X>1).

Jutta, what kind of use case do you have in mind, i.e. how complex are
your benchmarking scenarios?

Regards,

--ck

Re: Are there significant performance enhancements in 0.56.x to be expected soon or planned in the near future?

[Dennis Jacobfeuerborn]

On 01/09/2013 01:51 PM, Lachfeld, Jutta wrote:
> Hi all,
> 
> in expectation of better performance, we are just switching from CEPH version 0.48 to 0.56.1
> for comparisons between Hadoop with HDFS and Hadoop with CEPH FS.
> 
> We are now wondering whether there are currently any development activities 
> concerning further significant performance enhancements, 
> or whether further significant performance enhancements are already planned for the near future.
> 
> I would now be loath to start benchmarking with 0.56.1 and then, a month or so later, detect that there have been significant performance enhancements in CEPH in the meantime.

There shouldn't be any major changes since v0.56.x is a stable release and
as such should only receive bug-/securityfixes and non-risky improvements.
Any changes that would result in a significant change in performance would
probably be too disruptive for a stable release series.

Regards,
  Dennis

Re: Are there significant performance enhancements in 0.56.x to be expected soon or planned in the near future?

[Sage Weil]

On Wed, 9 Jan 2013, Dennis Jacobfeuerborn wrote:
> On 01/09/2013 01:51 PM, Lachfeld, Jutta wrote:
> > Hi all,
> > 
> > in expectation of better performance, we are just switching from CEPH version 0.48 to 0.56.1
> > for comparisons between Hadoop with HDFS and Hadoop with CEPH FS.
> > 
> > We are now wondering whether there are currently any development activities 
> > concerning further significant performance enhancements, 
> > or whether further significant performance enhancements are already planned for the near future.
> > 
> > I would now be loath to start benchmarking with 0.56.1 and then, a month or so later, detect that there have been significant performance enhancements in CEPH in the meantime.
> 
> There shouldn't be any major changes since v0.56.x is a stable release and
> as such should only receive bug-/securityfixes and non-risky improvements.
> Any changes that would result in a significant change in performance would
> probably be too disruptive for a stable release series.

That is generally true.  One exception is that there may be some simple 
changes that can decrease the impact of data migration on performance.  
There are some changes we made for a customer that seem to make a big 
difference and will be making it into the main tree (and hopefully 
bobtail, and possibly even argonaut) shortly.

sage

Re: Are there significant performance enhancements in 0.56.x to be expected soon or planned in the near future?

[Mark Kampe]

Performance work is always ongoing, but I am not aware of any
significant imminent enhancements.  We are just wrapping up an
investigation of the effects of various file system and I/O
options on different types of traffic, and the next major area
of focus will be RADOS Block Device and VMs over RBD.  This is
pretty far away from Hadoop and probably won't yield much fruit
until March.

There are a few people working on Hadoop integration, and I
have not been closely following their activities, but I do
not believe that any major performance work will be forthcoming
in the next few weeks

On 01/09/2013 04:51 AM, Lachfeld, Jutta wrote:
> Hi all,
>
> in expectation of better performance, we are just switching from CEPH version 0.48 to 0.56.1
> for comparisons between Hadoop with HDFS and Hadoop with CEPH FS.
>
> We are now wondering whether there are currently any development activities
> concerning further significant performance enhancements,
> or whether further significant performance enhancements are already planned for the near future.
>
> I would now be loath to start benchmarking with 0.56.1 and then, a month or so later, detect that there have been significant performance enhancements in CEPH in the meantime.

Re: Are there significant performance enhancements in 0.56.x to be expected soon or planned in the near future?

[Mark Nelson]

On 01/09/2013 06:51 AM, Lachfeld, Jutta wrote:
> Hi all,
>
> in expectation of better performance, we are just switching from CEPH version 0.48 to 0.56.1
> for comparisons between Hadoop with HDFS and Hadoop with CEPH FS.
>
> We are now wondering whether there are currently any development activities
> concerning further significant performance enhancements,
> or whether further significant performance enhancements are already planned for the near future.
>
> I would now be loath to start benchmarking with 0.56.1 and then, a month or so later, detect that there have been significant performance enhancements in CEPH in the meantime.
>
> Regards,
> Jutta.
>
> -
> jutta.lachfeld@ts.fujitsu.com, Fujitsu Technology Solutions PBG PDG ES&S SWE SOL 4, "Infrastructure Solutions", MchD 5B, Tel. ..49-89-3222-2705, Company Details: http://de.ts.fujitsu.com/imprint
> --
> 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
>

Hi Jutta,

As Wido mentioned there have been some performance improvements, 
especially with small IO sizes.  The conclusion section of the 
performance preview may be useful for you:

http://ceph.com/uncategorized/argonaut-vs-bobtail-performance-preview/

One oddity is that there may have been some regression for 128k reads. 
Overall though I'd say that performance has improved, especially on XFS.

I don't think it's likely we will be pushing any performance patches to 
the bobtail series, but it's possible performance could change as a 
result of a bug fix.

For what it's worth, I've started performing sweeps over ceph parameter 
spaces (and looking at underlying io schedulers) to see how tuning 
affects ceph performance under different scenarios.  I'm hoping to be 
able to release the results later this month.

Mark