Hi, in-line responses.



From: Yaniv Kaul <yk...@redhat.com>
Sent: Monday, August 7, 2017 7:42 AM
To: Moacir Ferreira
Cc: users@ovirt.org
Subject: Re: [ovirt-users] Good practices

On Sun, Aug 6, 2017 at 5:49 PM, Moacir Ferreira 
<moacirferre...@hotmail.com<mailto:moacirferre...@hotmail.com>> wrote:

I am willing to assemble a oVirt "pod", made of 3 servers, each with 2 CPU 
sockets of 12 cores, 256GB RAM, 7 HDD 10K, 1 SSD. The idea is to use GlusterFS 
to provide HA for the VMs. The 3 servers have a dual 40Gb NIC and a dual 10Gb 
NIC. So my intention is to create a loop like a server triangle using the 40Gb 
NICs for virtualization files (VMs .qcow2) access and to move VMs around the 
pod (east /west traffic) while using the 10Gb interfaces for giving services to 
the outside world (north/south traffic).

Very nice gear. How are you planning the network exactly? Without a switch, 
back-to-back? (sounds OK to me, just wanted to ensure this is what the 'dual' 
is used for). However, I'm unsure if you have the correct balance between the 
interface speeds (40g) and the disks (too many HDDs?).

Moacir: The idea is to have a very high performance network for the distributed 
file system and to prevent bottlenecks when we move one VM from a node to 
another. Using 40Gb NICs I can just connect the servers back-to-back. In this 
case I don't need the expensive 40Gb switch, I get very high speed and no 
contention between north/south traffic with east/west.

This said, my first question is: How should I deploy GlusterFS in such oVirt 
scenario? My questions are:

1 - Should I create 3 RAID (i.e.: RAID 5), one on each oVirt node, and then 
create a GlusterFS using them?

I would assume RAID 1 for the operating system (you don't want a single point 
of failure there?) and the rest JBODs. The SSD will be used for caching, I 
reckon? (I personally would add more SSDs instead of HDDs, but it does depend 
on the disk sizes and your space requirements.

Moacir: Yes, I agree that I need a RAID-1 for the OS. Now, generic JBOD or a 
JBOD assembled using RAID-5 "disks" created by the server's disk controller?

2 - Instead, should I create a JBOD array made of all server's disks?

3 - What is the best Gluster configuration to provide for HA while not 
consuming too much disk space?

Replica 2 + Arbiter sounds good to me.
Moacir: I agree, and that is what I am using.

4 - Does a oVirt hypervisor pod like I am planning to build, and the 
virtualization environment, benefits from tiering when using a SSD disk? And 
yes, will Gluster do it by default or I have to configure it to do so?

Yes, I believe using lvmcache is the best way to go.

Moacir: Are you sure? I say that because the qcow2 files will be quite big. So 
if tiering is "file based" the SSD would have to be very, very big unless 
Gluster tiering do it by "chunks of data".

At the bottom line, what is the good practice for using GlusterFS in small pods 
for enterprises?

Don't forget jumbo frames. libgfapi (coming hopefully in 4.1.5). Sharding 
(enabled out of the box if you use a hyper-converged setup via gdeploy).
Moacir: Yes! This is another reason to have separate networks for north/south 
and east/west. In that way I can use the standard MTU on the 10Gb NICs and 
jumbo frames on the file/move 40Gb NICs.


You opinion/feedback will be really appreciated!


Users mailing list

Users mailing list

Reply via email to