vSphere 5.5 Storage Enhancements Part 3: MSCS Updates
A number of new enhancements around Microsoft Clustering Services (MSCS) have been introduced in vSphere 5.5. I wanted to cover those in this post as I know many of you continue to use MSCS for service availability in your vSphere environments.
Round Robin Path Policy
In the past, round robin path policy could not be supported on shared storage in MSCS. This was due to the inability to release SCSI reservations on certain paths when the reservation was placed on a different path to the same disk. With the release of vSphere 5.5, this issue has now been addressed and we can support RR on storage that is being consumed by MSCS nodes.
Support for FCoE & iSCSI Protocols
In the past, only fiber channel protocol was supported for shared storage. Whilst there have been various conversations around the use of iSCSI initiators in the Guest OS, and limited support for FCoE (fiber channel over Ethernet), with the release of vSphere 5.5, we now fully support both FCoE & iSCSI protocols for shared storage access.
Support for Windows 2012 MSCS
A final enhancement to MSCS is support for Windows 2012 running as a Guest OS and clustered. I know a number of you have been looking for this.
Finally, in vSphere 5.1, we increased the number of supported MSCS nodes from two to five. All of the new vSphere 5.5 features included in this post apply all configurations, from two node up to five node.
Always refer to KB 1004617 and KB 1037959 for the latest support information on MSCS running in vSphere environments.
Some Opinion on MSCS
MSCS remains one of the very last use cases for Physical Mode Raw Device Mappings (pRDMs). The RDMs are required to allow SCSI reservations to be passed directly to the quorum/witness device without being handled/interpreted by the VMkernel I/O stack. You cannot use VMDKs in MSCS setups unless all nodes are on one host (but what’s the point of that?). However you need to be aware that using RDMs can limit your ability to use other vSphere features, and are a drag to deal with when doing a storage refresh. I’m sure many of you are aware of vSphere HA which can mitigate the use of MSCS is many cases. However HA will only restart a virtual machine and doesn’t really understand application availability. In vSphere 5.5, a new product called vSphere App HA can now provide availability for applications as well. This is built on top of our Hyperic technology and has the ability to restart an application without restarting the whole VM (though this is possible too if the application doesn’t respond accordingly). Worth checking out.
Great article Cormac. There are some excellent improvements in vSphere 5.5. For business critical apps that need failover clustering there are compelling reasons to make the shift to 5.5. It’s interesting to note that virtual mode RDM’s can also be used for some cases for clustering as per the KB and clustering guide, but pRDM’s are still the way to go. But I’m still looking forward to the day where PVSCSI can be used with failover clustering to enhance IO throughput and reduce latency and we don’t have to rely on LSI vSCSI adapters any longer.