In our environment we have been running ESX since just after version 3 was released running fiber channel to our storage array. We have has a couple different storage vendors (HP 3Par and Pure Storage) durring that time. In all that time nearly every major problem we have had has been related to storage. Often related to SCSI locks causing multiple hosts to become unresponsive. The problem often cascades to an outage as we try to work with support to determine the cause and get everything stable again.
My question is, what is a better solution? How can I have shared storage with all the advantages it brings but eliminate or greatly reduce the chance of my storage causing mulitiple hosts to check out/become unresponsive? Is iSCSI more stable, is a NFS share the way to go, or is Fiber Channel the best bet?
Please share your thoughts, this has become a great frustration for us.
Just my 2c. ...
I've seen a lot of environments, using all kinds of storage, ranging from NFS to Fiber, iSCSI ... And so on ...
Most of the environments never had any issues with shared storage, doesn't matter what technology was used.
The environment that did have issues are plenty too, but there was always some reason for it, wrong configuration, hardware or software problems, network issues, something.
It all comes down to really hunt for the cause, sometimes it's very difficult, sometimes it can be something like firmware of a hba not being up to date, conflicting ip-addresses or whatever. For some cases it took days if not weeks before the issue was found, depending on how frequent the issues occured or just how much resources were available to research the problem.
In short, I don't really have a preference, and all technologies have their own challenges when it comes to troubleshooting.
Welcome to the Community - I am sorry you have had these issues - Because shared storage is integral to a vSphere environment you have to make sure you are designing the storage appropriately - personally I have no issues ith FC SANs including 3PAR - the same goes for iSCSI and NAS/NFS -
As much as I don't like ever suggesting there's a "best" answer to these things, to suggest you can move from FC to iSCSI to improve stability is crazy.
Can you tell us more about your current platform? If you're STILL running ESX3 it won't be remotely fair to blame storage for your issues. I've run several different storage platforms from Dell's entry level iSCSI up to 3PAR FC and although failures can happen just about anywhere, I've never heard of "SCSI lock causing hosts to become unresponsive".