Hello,
I would like to set up our UCS infrastructure to handle correctly Netapp NFS storage in a vmware environment (5.5 U1).
First of all, our setup:
1x UCS Chassis 5108 with some C series blade servers (M3 + M4)
2x Fabric Interconnect 6248 with firmware 2.2(3h)
FI are connected to two distribution routers (Cisco 6500) in switched mode with two 10GBit-SR per FI (total 4 uplinks)
Our Netapp OnTap (8 Cluser Mode) 8040 is connected with 4 uplinks the same way to the two Cisco 6500. It expose the NFS lun via one Storage Virtual Machine with one IP.
On UCS I've provisioned (via template) services profiles. The most important part is that any network service (i.e. management network, production network, backup network, storage network) is created twice, setting for every vNic as fabric-id, fabric A or fabric B. In this way, in example, I've for the storage network two nic as follow:
- NFS-A (via Fabric A)
- NFS-B (via Fabric B)
(this for every service).
So, from a vmware point of view, focusing only on the storage path, I've for every host one vSwitch with two network adapters (jumbo frames enabled and vaai Netapp + plugin (1.01-21) installed and working). The vSwitch has one vmkernel port to connect to the SVM and mount then the volume.
My first question is:
- it's correct to set the adapters in active/active (default load balancing) for the storage traffic or should I use another setup?
Is there a best practice to follow?
Thanks a lot!
Simon
