4 nodes ceph configuration

[kaltsi]

Hi,

I would like build a 4 nodes cluster system with shared filesystem using ceph.
All nodes are the same. 48 core Intel CPU, 2x 100GB SFP network, 4x 1T normal disk, 256GB memory
I don't need high IO speed, but the stable state is important.
I going to create a raid5 storage from the 4 disk with 2 volumes. One of these volume will about 200GB for OS and the another will be the shared storage maybe 2.5TB.

I would like use template systems with linked clone. All templates and VM will located the shared storage area.
All VMs will have snapshots or will be a new template. Therefore I choused the caph.

Is it possible to use ceph and share this about 2.5TB disk spase between the nodes?
Could you please give me additional instruction for this system?

 

[PigLover]

Yes, you can.

But you might be better off not doing the Raid. Either let use one drive for OS/Local storage and let Ceph have the other three 1TB drives directly as OSDs or - better - get a small SSD for OS/Local storage and let Ceph manage all four drives directly as separate OSDs.

The reason is two-fold. First, by using the Radi5 and then Ceph you are doing "redundancy on top of redundancy", which adds no additional resiliency to your system and wastes space for no real value add.

Secondly, Ceph works better with more OSDs to manage. You'll get better IO performance with more OSDs to schedule IO onto. And it will perform better if/when you have disk faults.

 

[kaltsi]

Hi

Thank you for your detailed answers.
But I have to use what I have. I cannot increase the system.

But you are right. I will have double redundancy.
What do you think about this solution? I going to create 2x raid0 2-2 disk with more volumes? In this case I can use zfs for OS to be mirrored and I will have more disk for ceph.
It is a lutke bit strange think.