tmac> Depends on how many nodes you have.
Heh, two? I've got two clusters, and a bunch of standalone filers, all running 7.3.x. When looking at upgrade options, it does look like moving to 8 with vfilers might be a good thing to do, esp if it would let us get our FC heavy filers seperatated from our NFS only filers.
I need to watch that introduction from Netapp at some point...
tmac> They reserve port 13-20 (8 ports) as an ISL. tmac> That allows you to connect up to 12 nodes. tmac> There is an expansion slot on the 5010 to allow tmac> for more ISL's and more nodes, if needed. I think they take you to tmac> a set of 5020's for a full 24-node cluster.
tmac> Oh, and you may think it seems excessive, it is if you only have tmac> a handful of nodes. But, if you are purchasing from NetApp, tmac> they basically, "throw in" the extra switches. i.e. the 5010's tmac> and the 2960's
Ok, that makes more sense that these switches are included in the basic cluster setup. It just seemed excessive for a two or three node cluster that I'd be likely to have in my environment.
tmac> SFO - Storage failover tmac> CFO - Controller Failover
tmac> CFO is used when you have a two node system. It is basically the tmac> same thing as failover in 7-mode/ONTAP7
tmac> Storage Failover allows for only aggregates to failover, not tmac> necessarily waiting on the controller. Since there is a global tmac> namespace and logical interfaces, controller failover is not tmac> needed when using more than 2 nodes.
How well does cluster mode work with CIFS, iSCSI and FC failovers? I know NFS failovers will be just fine since they seem to work ok with OnTap 7. But I worry about the block based modes more.
tmac> And as been said a few times now...you get no SFO/CFO capability tmac> in the simulator. they are only stand alone nodes.
Bummers... hard to test how things would work in SFO/CFO failures ahead of time.