[Openstack] Why GlusterFS should not be integrated with OpenStack

John Mark Walker johnmark at johnmark.org
Tue Sep 10 14:42:58 UTC 2013


Hi there,

On Tue, Sep 10, 2013 at 10:28 AM, Maciej GaƂkiewicz
<macias at shellycloud.com>wrote:

>
> I am also using glusterfs but for different purposes. I hope to
> replace it with cephfs when it will be production ready. What I
> dislike in Glusterfs is it's stability. Many times it was not working
> as expected during brick replacement which is crucial for me. I am
> taking about version 3.3.0 or previous. Management of many glusterfs
> volumes generated many problems for me:( I am looking forward to test
> 3.4.0.
>
>
I'm happy to help you test out 3.4. A lot has changed with it, and much of
it affects the use cases you described, specifically libgfapi, the QEMU
integration, Nova integration, and the block device translator.

Our dev process is now also much cleaner and inclusive - you can see the
results of our recent 3.5 feature decisions here:
http://www.gluster.org/community/documentation/index.php/Planning35

Incidentally, you seem to be allergic to qcow2 files :)  We have some
performance numbers that may be eye-opening to you (and many others).

As Jeff mentioned in his response, there are many benefits from a unified
storage backend with file semantics that doesn't silo your data. Let me
know if I can be of any help.

-John Mark Walker
Gluster Community Leader
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstack.org/pipermail/openstack/attachments/20130910/dcea157f/attachment.html>


More information about the Openstack mailing list