Open Stack

I always preferred RAID5 over JBOD myself, but you always need to bench first and decide after :)

Razique Mahroua - Nuage & Co
razique.mahroua at gmail.com
Tel : +33 9 72 37 94 15



Le 16 janv. 2013 à 18:11, Gui Maluf <guimalufb at gmail.com> a écrit :

> Hey Sean,
> Did you found which is the best set up for your RAID?
> I'm facing the same situation here.
>  I've 11 dell r710 servers, with 4x3TB disk each, and I'm wondering if worth it using a RAID5 or JBOD for some server that will run a compute-network-volume node.
> Did you got any solution?
> Thanks in advance
> 
> 
> On Sat, Jan 12, 2013 at 7:21 PM, Razique Mahroua <razique.mahroua at gmail.com> wrote:
> Sorry, the first sentence is wrong : 
>> Hi,
>> actually, even for a raid 10 you would end up with 3TB usable (more around 2800 GB)
> In both cases you end up with 2 TB :)
> 
> Regards,
> Razique Mahroua - Nuage & Co
> razique.mahroua at gmail.com
> Tel : +33 9 72 37 94 15
> 
> <NUAGECO-LOGO-Fblan_petit.jpg>
> 
> Le 12 janv. 2013 à 22:19, Razique Mahroua <razique.mahroua at gmail.com> a écrit :
> 
>> Hi,
>> actually, even for a raid 10 you would end up with 3TB usable (more around 2800 GB). You could even go for either a RAID6 or RAID10 actually, since you have 4TB you would end up with 2TB in both cases
>> 
>> Performance wise RAID10 is in most cases faster when it comes to reading speed, sometimes slower when it comes to writing speed, it all depends on you controller. You would bench both RAIDs and pick the faster one :)
>> 
>> 
>> Razique Mahroua - Nuage & Co
>> razique.mahroua at gmail.com
>> Tel : +33 9 72 37 94 15
>> 
>> <NUAGECO-LOGO-Fblan_petit.jpg>
>> 
>> Le 11 janv. 2013 à 23:23, Sean Bigdatafun <sean.bigdatafun at gmail.com> a écrit :
>> 
>>> Razique,
>>> 
>>> Let's think of the following scenario, we use Nova to build an EC2-kind of compute cluster. For EC2, they offer 
>>> 
>>> M1.Small = 1 vCPU + 1.7GB Mem + 160GB Disk
>>> M1.Medium = 2vCPU + 3.75GB Mem + 410GB Disk
>>> M1.Large = 4vCPU + 7.5GB Mem + 850GB Disk
>>> M1.ExtraLarge = 8.vCPU + 15GB Mem + 1.7TB Dsik
>>> ...
>>> 
>>> This means that they at least have 1.7TB disk space on a physical server (presumably 2 x 1TB disk).  
>>> 
>>> 
>>> What RAID option  would we use in designing our cluster? For our Nova compute nodes,
>>> 
>>> -- Do we have a JBOD configuration of 2 disks?
>>> -- do we have a RAID 0 of 2 disks?
>>> -- or do we use more enterprise like RAID 5?
>>> 
>>> 
>>> Thanks and hopefully this starts some discussion in this direction.
>>> 
>>> Thanks,
>>> Sean
>>> 
>>> On Fri, Jan 11, 2013 at 12:46 AM, Razique Mahroua <razique.mahroua at gmail.com> wrote:
>>> Hi Sean, 
>>> I may be wrong, but I think there is not any "best" RAID strategy when it comes to the disk for the nodes or a "normal" one. All RAID type have their pros and cons.
>>> It'a all about security over performance against total space that would end up usable. What RAID controller is that?
>>> 
>>> Regards,
>>> Razique Mahroua - Nuage & Co
>>> razique.mahroua at gmail.com
>>> Tel : +33 9 72 37 94 15
>>> 
>>> <NUAGECO-LOGO-Fblan_petit.jpg>
>>> 
>>> Le 11 janv. 2013 à 08:36, Sean Bigdatafun <sean.bigdatafun at gmail.com> a écrit :
>>> 
>>>> When I have a set of rack servers that is 1TB * 4 disks in each of it as Nova Compute Node, what is the best RAID (or JBOD) strategy OpenStack community recommend? 
>>>> 
>>>> Basically I'd like to know what a normal disk setup a compute node should be using.
>>>> 
>>>> Thanks,
>>>> Sean
>>>> _______________________________________________
>>>> Mailing list: https://launchpad.net/~openstack
>>>> Post to     : openstack at lists.launchpad.net
>>>> Unsubscribe : https://launchpad.net/~openstack
>>>> More help   : https://help.launchpad.net/ListHelp
>>> 
>>> 
>>> 
>>> 
>>> -- 
>>> --Sean
>>>  
>> 
> 
> 
> _______________________________________________
> Mailing list: https://launchpad.net/~openstack
> Post to     : openstack at lists.launchpad.net
> Unsubscribe : https://launchpad.net/~openstack
> More help   : https://help.launchpad.net/ListHelp
> 
> 
> 
> 
> -- 
> guilherme \n
> \t maluf

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstack.org/pipermail/openstack/attachments/20130116/40123382/attachment.html>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: NUAGECO-LOGO_logo_complet.gif
Type: image/gif
Size: 9524 bytes
Desc: not available
URL: <http://lists.openstack.org/pipermail/openstack/attachments/20130116/40123382/attachment.gif>