Had a query from a Partner that I can't find an answer to, this is it:
At the moment we have 4 Ethernet ports on the filer aggregated in a Virtual Interface “eth0” with IP 192.168.101.112.
On the filer we have 3 volumes, with 3 NFS exports.
We would like to give each NFS export its own IP address, all using the “eth0” interface.
So one export would be 192.168.101.113.
Next one would be 192.168.101.114.
Next would be 192.168.101.115.
This would allow us to utilize the “Route based on IP hash” aggregation on our VMware machines, and allow us to spread VMware traffic across multiple interfaces on the VMware nodes.
At the moment, because all the NFS exports have the same IP address, VMware routes the traffic across one interface – limiting us to 1Gbit. We would rather have 1Gbit for each volume on the Netapp.
On our Bluearc system we simply create an “EVS” with its own IP address, and this IP is then exposed on the 6 port aggregate we have on it.
We then assign a volume to the EVS, and create the NFS export. We would create an EVS for each volume and have an export on each.
Basically, we would like to achieve the same on the Netapp filer – multiple IP addresses – one for each NFS export – all sharing the same aggregate. This would allow us to greatly improve the performance of our VMware on the Netapp filer (which is currently limited to 1Gbit throughput across 4 Ethernet interfaces and 3 volumes) to somewhere near what we achieve on our Bluearc system (which we get somewhere between 2-3Gbit throughput across 3 Ethernet interfaces and 3 volumes.)
Now I have seen this whitepaper which talks about the load balancing method mentioned, but does not achieve what is being asked here. Does anyone know if it is possible or if he could go about the same thing in a different way?