[Gluster-users] Build disk tiering offer with GlusterFS

[Hugues Lepesant]

Calors,

 
Thank for your answer.

 
The GlusterFS will be presented using NFS through a 10g network to a farm of KVM servers managed by Cloudstack.

I need space (1T hdd) and redundancy (GlusterFS) and want to have two disk offer (standard and faster)

 
hug
 
-----Message initial-----
De:Carlos Capriotti <capriotti.carlos at gmail.com>
Envoyé:mar. 25-03-2014 19:10
Sujet:Re: [Gluster-users] Build disk tiering offer with GlusterFS
À:Hugues Lepesant <hugues at lepesant.com>; 
Cc:gluster-users at gluster.org; 
 
Hugues:
 I tend to agree with you with your disk setting for system. I have that myself and it is rock solid, and lightning fast. 
 For your 11 x 1 To (tera octets, uh ? The French way), I'd tend to agree, because these 7.2 Krpm disks feel a bit slow, but for the 11x600 GB 15 Krpm... I think that will be a waste of diskspace. That is my IMPRESSION anyway, based on my similar 10x300 GB 10K rpm (PE2900 with PERC 5/i).  This setting can peak 800 MBytes/second, and my worst reading with it are around 400 MBytes/s.
 
 So, you can easily saturate regular network cards, in theory (1 Gbps), or at least, have a very decent data stream with 10 Gbps.
 My suggestion: install a quick test system with very basic tools on your 146 GB raid1 volume, just to be quick and do NOT include the data raid volumes.
 
 Next, make performance tests with each variation, RAID 5, RAID6 (if you have it), and RAID50, with both disk sets.
 Well, supposing of course you have the the hardware available, which sounds you have.
 
 KR, 
 Carlos


On Tue, Mar 25, 2014 at 4:30 PM, Hugues Lepesant <hugues at lepesant.com> wrote:
 

Hi all,

 
I can have Dell PowerEdge server with 3 groups of RAID drive.

 
1st : 2x146Go in RAID1 for the system

2nd : 11x1To 7.2K RPM in RAID50 (with one spare)

3rd : 11x600Go 15K RPM in RAID50 (with one spare)

 

 
Can I make a kind of tiering with two gluster volumes, one for each speed of disk, and  shared (by NFS) them on two IP address ?

 
Regards,

Hugues

 
 
_______________________________________________
 Gluster-users mailing list
 Gluster-users at gluster.org
 http://supercolony.gluster.org/mailman/listinfo/gluster-users
 
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://supercolony.gluster.org/pipermail/gluster-users/attachments/20140327/aacf5883/attachment.html>