Without boring you with the details, I'd like to run two servers from the same DAS. One server would be the primary server and would have read / write access to the DAS. The other server will be the backup and will have read-only access to the device until the primary server goes down, at which point it will gain write access. I'm hopeful that this solution could viably limit potential downtime on a mission critical alarming, control, and data acquisition system to fewer than 10 seconds if and when one of the servers finally gives out.
I've evaluated a number of different solutions, and I'm not super keen on any of the standard options. A failover cluster seems to be the most suitable option, but I'm still concerned that it won't provide 100% coverage.
EDIT: Apparently I'm going to bore y'all with the details. The servers aren't really serving much of anything, other than an occasional report served up by SSRS. I'm running a SCADA package on these servers that is constantly polling PLCs for data, logging some subset of those datapoints to a DB, and then rendering graphics that show the current status of a few dozen different machines. These servers will periodically dispatch alarms to pagers / cell phones / email addresses.
The bulk of the data on these machines would be functionally identical if I simply made a system image of one, restored it to the other, powered them up, and walked away. The information that we're really concerned about is tucked away in strange directories, or in some cases, exists in the registry. We need a solution that will keep system directories, hidden directories, and ideally even the registry either synchronized or shared between the two machines. People get re-assigned, they quit, they get hired, they get fired. Personnel on site consistently make changes to notification groups, the IT folks on site like to install antivirus software and remote access software and they periodically mess with the network settings.
It doesn't matter so much if the backup machine takes over and the numbers in its DB are all off by half a percent. But, if the primary machine goes down, and the backup machine can't connect to the network due to a configuration or software problem that was fixed on primary PC, then the client sees non-trivial down time. If the primary machine goes down, and the notification profiles in the backup machine are out of date, then the operators don't get their alarms and product gets damaged, production slows down, or god forbid, mechanical failure causes more serious damage.
Somehow, in light of this fact, this facility, like so many others, hasn't so much as connected a USB hard drive and set up a backup schedule on the pc that they have there now.