I would just like some clarity, in the active - active iscsi setup the recommendation is to bond nic's, does the balance-alb give bandwidth aggregation and failover?
I have two dual port 10GB cards in each server for iscsi and 1 10GB card for replication (direct connection). On nodeA can i bond the ports on card one and plug them into switch one into trunk ports (LAG Group ) and bond ports on card two and plug them into switch two trunk ports (LAG Group) and do the same for nodeB. Or do i need to plug one port on each card into each switch and trunk the two switches together with a LAG group.
To be honest I would not do any bonding (yes I know that guide that your looking at but there is another that is no bonds), I would use MPIO and dedicate the 1 x 10GbE NIC for the Volume Replication and the other 2 x 10GbE NICs for the Virtual IPs (keep them on separate networks or you will have routing issues).
I would have 4 x 10Gb ports on each server going to 2 x 10GB switches given me four virtual IP addresses is that correct which will giving me 2 VIP used for resources on node A and 2 VIP for resources on node B. Then i will enable MPIO on the VMware hosts that have 4x10GB Ports
I would then have 1 x 10GB port on each server for replication (direct link)
What i was hoping to do was that I have 4 vmware hosts split into two clusters (2 hosts each) So I am wanting to configure open-e Active - Active to use 2x10GB cards for cluster1 iscsi targets and 2x 10GB cards for cluster2 iscsi targets and 2 10xGB for replication. Does this make sense