Different implementations for File Server Failover Clustering

I have got this question many times in last few days and also seen a couple of implementations where Hyper-V host clustering was configured alongside File Server failover clustering on same hosts. The intention of this blog is to cover some failover clustering scenarios and reiterating the fact that it is not recommended to install any role in Hyper-V parent partition and if you have configured Hyper-V host clustering to run highly available virtual machines, make sure you do not install any other role in the Hyper-V parent partition and this applies to file server role too. The reason you should not install such a configuration is as you load the hyper-V parent partition by adding more roles, it may affect the performance of the hyper-V host and all the virtual machines running on Host. Check here , here and here where it clearly mentions that Microsoft Hyper-V Server parent partition is specifically for system management.

Second interesting question that comes up in discussion is how to configure File server clustering?

• Host file server failover clustering.
• Configure file server in highly available virtual machine.
• Configure guest file server clustering.
• Configure guest file server clustering on HA Virtual machines running on Hyper-V host clustering.

 I am not going to reinvent the wheel as Failover Clustering program manager Elden has already shared good thoughts on this topic here and shared below.

• Host file server failover clustering.

If host goes down…shares failover to another node ..if host needs to be patched..shares can be failed over to other node. This scenario does not provide you the benefits of virtualization but at same time it is preferable if there is considerably high IO happening on file server.

• Configure file server in highly available virtual machine.

HA VM can live migrate or failover from one node to another…for patching purposes we can leverage live migration for moving file server HA VM from one host to another…however during failover the file server VM will take couple of seconds to come online on another node and then server service and SMB shares will come online. Host based failover clustering also monitors the health of the VMs, so if the virtual OS crashes, hangs, or blue screens, it can be automatically restarted, however the end user would need to wait a little longer while the VM state is loaded and started. While Host Clustering can monitor the state of the virtual OS, it does not monitor the health of applications inside the VMs.

• Configure guest file server clustering.

You set up a highly available file server failover cluster across two virtual machines which are running on Hyper-V hosts. If one of the Hyper-V host goes unexpectedly down the guest file server cluster VM will turn off and shares will failover to other guest file server cluster VM running on other hyper-V host. In this scenario if we have to patch the hyper-v host we have to failover the shares from one VM to another.

• Configure guest file server clustering on HA Virtual machines running on Hyper-V host clustering.

Host+Guest clustering provides both Virtual machine [VM] and application mobility. You will be able to move the guest file server cluster virtual machine from one host to another using live migration without any downtime and cluster service in guest clustering will keep checking the health of server service and SMB shares and will failover the shares from one VM to another when needed. Though you have to be careful while planning Host+Guest clustering and make sure that you configure the anti affinity property for the guest cluster file server workloads so that they don’t preferably exist on same Hyper-V host node. You should also check mark the “auto start” property for HA VM so that when one of the physical host goes down and VM fails over to another node, it starts itself automatically so that file shares become available again. While during the failover as the VM will go down…the cluster service in guest clustering will failover the SMB shares to another HA VM running on another hyper-V host and will reduce the downtime.

 

There are some nice blog posts by failover cluster team and Jose Bareto which are nice reference.

http://blogs.msdn.com/b/clustering/archive/2008/06/21/8628515.aspx

http://blogs.technet.com/b/josebda/archive/2008/06/17/windows-server-2008-hyper-v-failover-clustering-options.aspx
Hope this blog will answer some of the questions in reference to implementation of file server failover clustering and advantages and disadvantages of each implementation solution. Thanks for your time and hope this was a reading pleasure.

GAURAV ANAND