DFS versus Fault Tolerance

[butlpau]

Hi,

Was hoping you could help please? We are running vSphere in a PoC with VDI at the moment and have 50 users up and running. We have the users Desktop and My Documents redirected to shares on a Windows 2008 64 bit server.

Recently, there was an issue with the Server in that it became almost unresponsive (incorrect version of Syamantec Endpoint was install - now resolved) but when this happenned the users could no longer access their Deskop files and their VDI session hung from their perception.

I am wondering how to get around this should it happen again?

I was looking at DFS but this would be costly in terms of storage required and if a second node was put in, it would potentially on aleviate the problem by 50%. Making the Server Fault Tolerant would work in theory but only if there was a hardware outage on the Server for whatever reason.

Then there is the option of setting up a Microsoft Cluster within the VMs but this is getting away from the whole virtualisation concept.

Unfortunatley we do not have the luxury of NFS shares and cannot not redirect the users Desktop and My Documents.

I was wondering if there is anything else I could implement that should the Server become unresponsive that it would not have any impact on the end users in this scenario?? Any comments appreciated.

Many Thanks,

Paul.

[1an3]

As is my understanding, If you have a fault-tolerant VM, all operating system actions are replicated to the secondary vm - so if your OS locks up in one (for whatever reason), it will lock up the other. FT is handy for when you might lose a ESX(i) host but don't want to lose the VM.

[golddiggie]

Is the server where the user desktop and documents are redirected to a VM or physical server? How many NIC's do you have allocated to that server? Have you thought about having more than one NIC in a load-balancing configuration? How many CPU's does the server have? What kind of disk is inside your storage (any additional details would be benefitial here), and how many are there (RAID level too)? View/VDI (it's really View now, not VDI with the latest releases) is heavily dependant on IOPs to get the job done. While only 50 users/systems shouldn't be an issue, you could be exceeding what your storage can handle. It could, also, be partially the LAN hardware in place. What are you running for the network from the desk all the way back to the servers? Make sure all switches are at the latest (stable) firmware version and settings are optimized for performance.

I would also call on resources from VMware since you're in a PoC phase. It's in their best interest to ensure the experience is a good one.

VCP4

[butlpau]

Thanks for the responses. The Desktops etc are redirected to a VM with provisioned storage from our production SAN. From the point of view of this topic, now I'm not really concerned what, if any, are the issues with the Windows (VM) Server hosting the shares, but more how to build some redundancy into the setup so if it does hang, the 'View' users are stuck without their Desktops etc... :smileyhappy:

Could I change my configuration different to handle a redundancy option that I don't seem to have now?

[golddiggie]

If performance is back to where it's within tolerance, then you'll just need to be super-vigilant when it comes to maintaining that server. Make sure you don't allow any items to get out of date that have the potential to impact it's performance. I would also build as much redundancy into that server as possible. Make sure the files it's sharing are backed up as often as possible. It would probably be a good idea to reach out to some other organizations that have gone through the PoC phase, and went to production with View and find out how they configured everything to ensure zero down time/data loss to the users. I'm sure that your VMware rep can get you that information so that you can talk with someone tha'ts a close match to your organization structure/size.

VCP4