[Openstack-operators] Scalable Glance Nodes

Narayan Desai narayan.desai at gmail.com
Tue Mar 19 12:07:28 UTC 2013


Yeah, we're using pacemaker for all of that.
 -nld


On Mon, Mar 18, 2013 at 10:39 PM, Joe Topjian <joe.topjian at cybera.ca> wrote:

> Cool - thank you very much!
>
> One more question: are you doing any HA for the Glace database?
>
>
> On Mon, Mar 18, 2013 at 9:43 PM, Narayan Desai <narayan.desai at gmail.com>wrote:
>
>> We're currently running two systems (in the middle of a migration). In
>> both, we just have the basic brick level replication, using 2 way
>> replication in one case, and 4 way in the other. We're running a relatively
>> small image archive; about 1TB.
>>
>> Because the machine has a decent network (QDR infinband using IPOIB), we
>> don't actually need multiple glance servers for scalability reasons, we
>> just have the services HA'd with pacemaker.
>>
>> Generally speaking, our workload tends to be relatively synchronized; at
>> periods of high use, you might have a single user firing up a hundred
>> instances. That leaves us network bound, but the infiniband tends to stand
>> up to that load pretty well.
>>  -nld
>>
>>
>> On Mon, Mar 18, 2013 at 8:55 PM, Joe Topjian <joe.topjian at cybera.ca>wrote:
>>
>>> Hi Narayan,
>>>
>>> If you don't mind, I'd love to know some details about your
>>> implementation:
>>>
>>> What type of replication are you using?
>>> How large is your Glance image repository and how many Glance servers
>>> are you using?
>>> Are you load balancing between Glance servers? And if so, what method?
>>> (DNS rr?)
>>> How much activity does your Glance service see?
>>>
>>> Thanks,
>>> Joe
>>>
>>>
>>>
>>> On Mon, Mar 18, 2013 at 8:35 PM, Narayan Desai <narayan.desai at gmail.com>wrote:
>>>
>>>> We've happily been using glusterfs with N way replication for the last
>>>> year for this, with no problems.
>>>>  -nld
>>>>
>>>>
>>>> On Mon, Mar 18, 2013 at 5:12 PM, Ahmad Ahmadi <ahmadidamha at yahoo.com>wrote:
>>>>
>>>>> Hi,
>>>>>
>>>>> How can I have multiple glance nodes in a way that, when I create an
>>>>> image from the controler, it would be coppied to all of the glance nodes?
>>>>> Now, when I define multiple glance endpoints in the controler
>>>>> (keystone), the images would only go to the first glance node.
>>>>>
>>>>> I also want to avoid putting the name of all glance nodes in the nova
>>>>> nodes.
>>>>>
>>>>> Thank you,
>>>>> Ad
>>>>>
>>>>> _______________________________________________
>>>>> OpenStack-operators mailing list
>>>>> OpenStack-operators at lists.openstack.org
>>>>> http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-operators
>>>>>
>>>>>
>>>>
>>>> _______________________________________________
>>>> OpenStack-operators mailing list
>>>> OpenStack-operators at lists.openstack.org
>>>> http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-operators
>>>>
>>>>
>>>
>>>
>>> --
>>> Joe Topjian
>>> Systems Administrator
>>> Cybera Inc.
>>>
>>> www.cybera.ca
>>>
>>> Cybera is a not-for-profit organization that works to spur and support
>>> innovation, for the economic benefit of Alberta, through the use
>>> of cyberinfrastructure.
>>>
>>> _______________________________________________
>>> OpenStack-operators mailing list
>>> OpenStack-operators at lists.openstack.org
>>> http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-operators
>>>
>>>
>>
>
>
> --
> Joe Topjian
> Systems Administrator
> Cybera Inc.
>
> www.cybera.ca
>
> Cybera is a not-for-profit organization that works to spur and support
> innovation, for the economic benefit of Alberta, through the use
> of cyberinfrastructure.
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstack.org/pipermail/openstack-operators/attachments/20130319/f74dfd56/attachment.html>


More information about the OpenStack-operators mailing list