Simple VMware HA (vSphere 4) with NAS

[datnus]

Hello all,

I would like to build a simple Vmware HA (vSphere 4) with NAS.

Let's say I use:

+ 2 servers

+ 2 switches

+ 2 NAS boxes.

Each server connects to 2 switches. Each NAS box connects to 2 switches, too.

As I understand, Vmware HA would use shared storage (NAS in this case) to store the vm files. VMware will reboot a VM into another server host while the vm files are still at the old shared storage.

1) Problems: What if the NAS fails?

Could I use 2 NAS boxes to mitigate the problem? I could let 2 NAS boxes to sync with each other.

But then how Vmware will fail over among the two NAS boxes?

2) Ppl love to use SAN. But except the performance, SAN is also a single point of failure, right?

Thank you very much for any help

[NuggetGTR]

1) Problems: What if the NAS fails?

Unfortunately you vm(s) go off with it.

with 2 NAS boxes you will still have an outage if one was to fail but if you were backing up on the second you could bring the vm online pretty quickly. But your NAS should have some form or redundancy like raid 5 + 1 or something so you should not get a total failure.

2) Ppl love to use SAN. But except the performance, SAN is also a single point of failure, right?

Yes SANs are also a single point of failure but again, the chances of a Datastore failure on a enterprise level SAN is almost a non event as there is so much redundancy built into them. I usually come across fabric switches having issues and may drop some paths, but then again you have redundant paths and backup HBA's and when I have lost the paths for an entire server(More likely then a SAN failure) HA will bring up all the machine in that server on the other servers in the cluster.

[datnus]

Thanks NuggetGTR. :smileygrin:

Is it possible to use vStorage multipath to failover from failed NAS to another healthy one?

Thanks

[NuggetGTR]

No, multipathing is to get more bandwith through to and from your storage as well as reduntancy but your vmdk file is sitting on one NAS if that was the fail if doesnt bring it up on another NAS or storage device, unless your using something like site recovery manager or plate spin something that keeps almost live copies of your virtuals incase of a DR situation.

what I can see working is something like EMC's VPLEX i think its called where it is actually doing virtual LUNs so you can move the storage around live and the ESX hosts dont know any difference its like vms but from a storage end. This is really top teir storage and i dont know many places running this yet, in Australia at least.