Hello All, I am about to deploy a new Ovirt platform, the platform will consist 4 Ovirt nodes including management, all servers nodes and storage will have the following config: *nodes server* 4x10G ports network cards 2x10G will be used for VM network. 2x10G will be used for storage connection 2x1Ge 1xGe for nodes management *Storage *4x10G ports network cards 3 x10G for NFS storage mount Ovirt nodes Now given above network configuration layout, what is best practices in terms of nodes for storage NFS connection, throughput and path resilience suggested to use First option each node 2x 10G lacp and on storage side 3x10G lacp?

The second option creates 3 VLAN's assign each node on that 3 VLAN's across 2 nic, and on storage, side assigns 3 nice across 3 VLANs?

-- Tal Bar-or _______________________________________________ Users mailing list Users(a)ovirt.org http://lists.ovirt.org/mailman/listinfo/users

Hello Tal It seems you have a very big overkill on your environment. I would say that normally 2 x 10Gb interfaces can do A LOT for nodes with proper redundancy. Just creating Vlans you can separate traffic and apply, if necessary, QoS per Vlan to guarantee which one is more priority. If you have 2 x 10Gb in a LACP 802.3ad Aggregation in theory you can do 20Gbps of aggregated traffic. If you have 10Gb of constant storage traffic it is already huge, so I normally consider that Storage will not go over a few Gbps and VMs another few Gb which fit perfectly within even 10Gb The only exception I would make is if you have a very intensive (and I am not talking about IOPS, but throughput) from your storage then may be worth to have 2 x 10Gb for Storage and 2 x 10Gb for all other networks (Managment, VMs Traffic, Migration(with cap on traffic), etc). Regards Fernando 2018-03-21 16:41 GMT-03:00 Yaniv Kaul <ykaul(a)redhat.com&gt;: > > > On Wed, Mar 21, 2018 at 12:41 PM, Tal Bar-Or <tbaror(a)gmail.com&gt; wrote: > >> Hello All, >> >> I am about to deploy a new Ovirt platform, the platform will consist 4 >> Ovirt nodes including management, all servers nodes and storage will have >> the following config: >> >> *nodes server* >> 4x10G ports network cards >> 2x10G will be used for VM network. >> 2x10G will be used for storage connection >> 2x1Ge 1xGe for nodes management >> >> >> *Storage *4x10G ports network cards >> 3 x10G for NFS storage mount Ovirt nodes >> >> Now given above network configuration layout, what is best practices in >> terms of nodes for storage NFS connection, throughput and path resilience >> suggested to use >> First option each node 2x 10G lacp and on storage side 3x10G lacp? >> > > I'm not sure how you'd get more throughout than you can get in a single > physical link. You will get redundancy. > > Of course, on the storage side you might benefit from multiple bonded > interfaces. > > >> The second option creates 3 VLAN's assign each node on that 3 VLAN's >> across 2 nic, and on storage, side assigns 3 nice across 3 VLANs? >> > > Interesting - but I assume it'll still stick to a single physical link. > Y. > > Thanks >> >> >> >> >> >> -- >> Tal Bar-or >> >> _______________________________________________ >> Users mailing list >> Users(a)ovirt.org >> http://lists.ovirt.org/mailman/listinfo/users >> >> > > _______________________________________________ > Users mailing list > Users(a)ovirt.org > http://lists.ovirt.org/mailman/listinfo/users > >