[pve-devel] manually add servers to the PVE corosync cluster

[Michael Svendsen]

Hi Stefan & co.,

Sounds good that you are testing both. I have been thinking on doing some
performance/scalability tests as well.
I have a handful Proxmox servers (not in cluster and not with shared storage
available)

How is your setup? Have you installed Ceph/Sheepdog on the Proxmox nodes
directly? Or do you run it on dedicated storage servers?
Also I would like to hear if you have installed Ceph/Sheepdog on servers
with also RAID configured, or you let Ceph/Sheepdog handle individual disks
directly?

Looking forward to hear your performance conclusions :-)



- Michael


-----Oprindelig meddelelse-----
Fra: pve-devel-bounces at pve.proxmox.com
[mailto:pve-devel-bounces at pve.proxmox.com] På vegne af Stefan Priebe -
Profihost AG
Sendt: 2. juli 2012 15:24
Til: pve-devel at pve.proxmox.com
Emne: Re: [pve-devel] manually add servers to the PVE corosync cluster


Am 29.06.2012 17:49, schrieb Martin Maurer:
>> Cc: pve-devel at pve.proxmox.com
>> Subject: Re: [pve-devel] manually add servers to the PVE corosync 
>> cluster
>>
>> I'm sorry it works even when the proxmox server is not in the same 
>> corosync cluster.
>>
>
> I am quite interested in your comparison between sheepdog and ceph, please
share your findings here.

Right now Alexandre and myself do heavy testing of both. But it seems that
both have it strength and weakness. Ceph has good squential write values and
seem to be really stable (if youn setup it correctly). Also it is really
good documented instead of sheepdog.

Sheepdog has good random I/O values but lacks in stability, documentation
and sequential stuff.

Greets
Stefan
_______________________________________________
pve-devel mailing list
pve-devel at pve.proxmox.com
http://pve.proxmox.com/cgi-bin/mailman/listinfo/pve-devel