Gluster users - Jan 2013 - Fw: performance evaluation of distributed storage systems

----- Forwarded by Sogand Shirinbab/Staff/BTH on 2013-01-22 10:55 -----

From:   Sogand Shirinbab/Staff/BTH
To:     gluster-users at gluster.org
Date:   2013-01-16 14:35
Subject:        performance evaluation of distributed storage systems



Hi,

I'm a phd student and as a part of my research I've compared performance
of different distributed storage systems (Gluster, Openstack, Compuverde). 
I would like you as an expert in your product to give me feedback on my 
work. What do you think about the way we've setup the system? does it 
affects the Gluster performance?

please find my paper as attachment to this mail!

Best Regards,
SogandShirinbab


Blekinge Tekniska H?gskolan
371 79 Karlskrona 
0455 -385709 

-------------- next part --------------
An HTML attachment was scrubbed...
URL:
<http://supercolony.gluster.org/pipermail/gluster-users/attachments/20130122/21bf192d/attachment.html>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: Performance Evaluation of Distributed Storage Systems-final.pdf
Type: application/octet-stream
Size: 2402131 bytes
Desc: not available
URL:
<http://supercolony.gluster.org/pipermail/gluster-users/attachments/20130122/21bf192d/attachment.obj>

I didn't see anything in the paper that mentioned the specs on your
load generating clients? Where can I get your test scripts for the
structured filesystem tests other than specsfs2k8? I assume you used
distributed + replicated with gluster? Is compuverde using a similar
algorithm or is it striping files? You mentioned using NFS/CIFS. NFS &
CIFS can have different performance characteristics so were the graph
results from NFS or CIFS and what versions, mounting protocols, etc?

Found this interesting tidbit in wiki :

January 2012 Compuverde, Blekinge Institute of Technology and Ericsson
received recognition from the Development of Knowledge and Competence
(KK-stiftelsen) in Sweden for a joint venture project on big data
storage solutions and cloud computing.

:)

On Tue, Jan 22, 2013 at 3:55 AM, Sogand Shirinbab
<sogand.shirinbab at bth.se> wrote:
> ----- Forwarded by Sogand Shirinbab/Staff/BTH on 2013-01-22 10:55 -----
>
> From:        Sogand Shirinbab/Staff/BTH
> To:        gluster-users at gluster.org
> Date:        2013-01-16 14:35
> Subject:        performance evaluation of distributed storage systems
> ________________________________
>
>
>
> Hi,
>
> I'm a phd student and as a part of my research I've compared
performance of
> different distributed storage systems (Gluster, Openstack, Compuverde). I
> would like you as an expert in your product to give me feedback on my work.
> What do you think about the way we've setup the system? does it affects
the
> Gluster performance?
>
> please find my paper as attachment to this mail!
>
> Best Regards,
> SogandShirinbab
>
>
> Blekinge Tekniska H?gskolan
> 371 79 Karlskrona
> 0455 -385709
>
>
> _______________________________________________
> Gluster-users mailing list
> Gluster-users at gluster.org
> http://supercolony.gluster.org/mailman/listinfo/gluster-users

>    I'm a phd student and as a part of my research I've compared
>    performance of different distributed storage systems (Gluster,
>    Openstack, Compuverde).
Compuverde? That's new to me. Oh wow.

"Software defined storage just got 400 % more efficient."

"Compuverde Gateway read and writes structured data 250 % more
efficient(sic) than well-known market competitors."

"The Compuverde software will help businesses lower their energy costs
with(sic) up to 50%"

I have not seen such a comprehensive pile of BS for a long time - thank you
for making my day.

The 400% claim?
http://compuverde.com/object-store/performance/
Vaguely-described tests against an unspecified competitor.

Finally:

"The Compuverde Object Store software features patented and patent-pending
technology"

The PHB will be impressed.

Regarding the PDF evaluation: I only skimmed it, but what exact version of
glusterfs did you use?  There have been a lot of changes between 3.2.5 and
3.3.0 for example.  Also:

"3.1.4 Gluster
...
The communication protocol between the load generating clients and the proxy
servers is NFS/CIFS"

So which was it? And what are "proxy servers" in the context of a
Gluster
test?

Gluster supports NFS exporting natively from one brick. CIFS requires using
Samba to re-export a glusterfs mount. However I would have thought using the
native glusterfs FUSE client would be a fairer test.

Regards,

Brian.

On 01/22/2013 04:55 AM, Sogand Shirinbab wrote:
> I'm a phd student and as a part of my research I've compared
performance
> of different distributed storage systems (Gluster, Openstack,
> Compuverde). I would like you as an expert in your product to give me
> feedback on my work. What do you think about the way we've setup the
> system? does it affects the Gluster performance?
Interesting work.  Some questions:

(1) Why was Ceph excluded?  It's readily available and has an almost 
identical object-storage interface, but no reason is given for leaving 
it out.

(2) The Atom/4GB/GigE storage nodes seem like an odd choice.  Have you 
tried testing on other kinds of platforms?

(3) More information on software versions and configurations would be 
very helpful.  For GlusterFS, there are significant differences both 
between versions and between different ways of organizing a volume 
across 384 disks.

(4) What program(s) other than SPECsfs did you use to generate load? 
The results for most of the tests seem very inconsistent with those for 
SPECsfs, especially for GlusterFS.

(5) How POSIX-compliant is the structured form of Compuverde?  Does it 
have full and proper support for things like fsync/O_SYNC, extended 
attributes, or atomic cross-directory rename?  Does it use FUSE, or 
interface to the system in some other way?

(6) Several of the statements made in section 5.3 seem inaccurate. 
GlusterFS only uses rsync for remote replication, which it seems clear 
from the rest of the paper would be irrelevant to these tests.  Also, 
it's not generally true that self-heal would actually be done from the 
proxy servers (though it could be initiated from there).  Lastly, this 
is one of the areas where version/configuration differences would make a 
huge difference in the results.

(7) The nature of the relationship between BTH and Compuverde needs to 
be more explicit.  Did it include configuration/tuning help?  Bug fixes? 
  Equipment loans or other material support?  If not, then it's just a 
matter of applying usual disclosure standards.  If so, then perhaps 
representatives from other projects (OpenStack as well) should have the 
same opportunity to make sure the results are representative of current 
best practices.

As you come from BTH, I believe your test result is biased, because I get
?2011 Establishing research agreement with BTH and Ericsson? from
compuverde.com.


As I could not even find a trial version of compuverde on compuverde.com, none
of us could help you verify your test result.

your email looks like a ad. , that`s bad.

Jules Wang.

At 2013-01-22 17:55:13,"Sogand Shirinbab" <sogand.shirinbab at
bth.se> wrote:
----- Forwarded by Sogand Shirinbab/Staff/BTH on 2013-01-22 10:55 -----

From:        Sogand Shirinbab/Staff/BTH
To:        gluster-users at gluster.org
Date:        2013-01-16 14:35
Subject:        performance evaluation of distributed storage systems




Hi,

I'm a phd student and as a part of my research I've compared performance
of different distributed storage systems (Gluster, Openstack, Compuverde). I
would like you as an expert in your product to give me feedback on my work. What
do you think about the way we've setup the system? does it affects the
Gluster performance?

please find my paper as attachment to this mail!

Best Regards,
SogandShirinbab


Blekinge Tekniska H?gskolan
371 79 Karlskrona
0455 -385709
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
<http://supercolony.gluster.org/pipermail/gluster-users/attachments/20130123/351a1894/attachment.html>