Often there is a debate if we should put Citrix XenDesktop PVS vDisk on CIFS or SAN LUN. I like CIFS’s simplicity and SAN LUN’s resilience. Can we have both? Answer is SMB 3.
To verify I setup a lab with NetApp 4 node cluster and Xendesktop 7 . I put vDisk on three different locations and tested with LIF failover and storage node failover. During the LIF failover and storage failover, IOMeter continues to generate work load.
- vDisk hosted on SMB 2 CIFS share on NetApp ONTAP 8.1
- SMB 3 CIFS share on NetApp ONTAP 8.2
- iSCSI multipathing LUN on NetApp ONTAP 8.1.2
- CIFS LIF move
- Storage node failover
You must have failover group setup properly to do this procedure. Please see my blog on how to setup failover group.
- Use the network interface modify command to change the Status Admin of the LIF to down (offline):
network interface modify -vserver vs1 -lif lif1 -status-admin down
2. Use the network interface modify command to change the status admin of the LIF to up (online):
network interface modify -vserver vs1 -lif lif1 -status-admin up
When thousands of desktop users share one infrastructure, virtual desktop solution resilience is critical. You should verify Clustered ONTAP each node can successfully fail over to another node. This helps ensure that the system is configured correctly and that you can maintain access to data if a real failure occurs.
Storage node failover procedure:
- Check the failover status by entering the following command:
storage failover show
- Take over the node by its partner:
storage failover takeover -ofnode nodename
- Verify that failover was completed storage failover show command.
- Give back the storage to the original node:
storage failover giveback -ofnode nodename
Verify that giveback was completed by using the storage failover show-giveback command.
- Revert all LIFs back to their home nodes by entering the following command:
network interface revert *
Table below shows the result of test:
During the LIF migration, there is no impact on the virtual desktop. During the node failover, vDisk on CIFS SMB 2 causes CIFS Share hung and VM hung for 1-3 minutes depends how fast the failover finishes.
SMB 3 and iSCSI LUN has no impact on VMs during storage node failover.
CIFS is much easier to setup than multipathing SAN LUNs. Once setup, you can mount one CIFS share to multiple PVS servers. When you update the vDisk on a CIFS share, all the PVS servers get updated. CIFS SMB 3 brings simplicity and resilience to the solution.