Ceph on single 10GB

A

andrea68

Renowned Member
Jun 30, 2010
158
2
83
Hi, I have 3 nodes 2x10GB nic each
I will use proxmox cluster + VM traffic on one Nic and Ceph on the other.

will it be enough?

Tnx in advance...
 
If I will use gluster it is possible? It is less heavy for nic and CPU?
Can be corosync share same nic with gluster?
 
andrea68 said:
If I will use gluster it is possible? It is less heavy for nic and CPU?
Can be corosync share same nic with gluster?
Click to expand...
Never run corosync on shared links [1]. With that said, at least one link needs to be on a separate NIC port. The additional links [1] may be shared, but preferably not. This is especially important if HA [2] is used.

[1] https://pve.proxmox.com/pve-docs/pve-admin-guide.html#chapter_pvecm
[2] https://pve.proxmox.com/pve-docs/pve-admin-guide.html#chapter_ha_manager
 
Alwin said:
Never run corosync on shared links [1]. With that said, at least one link needs to be on a separate NIC port. The additional links [1] may be shared, but preferably not. This is especially important if HA [2] is used.

[1] https://pve.proxmox.com/pve-docs/pve-admin-guide.html#chapter_pvecm
[2] https://pve.proxmox.com/pve-docs/pve-admin-guide.html#chapter_ha_manager
Click to expand...


Yes, I have 2 Nic to handle on each of the three server.
1x1GB and 1x10GB

In case I can't add more, how can you advise to use this links?
 
Alwin said:
Add more network cards! Otherwise don't use HA.
Click to expand...

Ok.
For the moment I think I will use gluster only for fast MANUAL migration, without HA.
Or maybe ZFS replication I will check the smartest solution with this limited resources...

Thank you!!!
 
Well I managed to update to 2 x 10 GB lan...
In you opinion ceph running well on his own dedicated nic 10GB?
We have 3 nodes proxmox xeon 64GB ram 6 x 1TB SSD datacenter version each...
 
andrea68 said:
In you opinion ceph running well on his own dedicated nic 10GB?
Click to expand...
Yes/No. It all depends on the workload and resource usage of other services. So in short, you will need to test it.
 
Ok, now I have the project all set.

3 Nodes with 2x6core Xeon E-2276G, 64GB Ram each, 1x2 Nmve for proxmox os and 1x6 1TB SSD datacenter as storage (each).
How can i test the workload on this? There's some tools I need to use?

Actually I have proxmox cluster on a dedicated 10GB and a free 10GB (x3) for storage...

Considering this scenario I'm really undecided between Ceph and GlusterFS

Thanks in advance for all hints ...
 
Last edited:
You must log in or register to reply here.
Top Bottom