>Hi,
>Can you expand on the problems experienced. We are currently running a Private Cloud, but the clients want the pricing to be the same as AWS.
>

We tried running a MSSQL mirror with witness setup in the cloud and 3 load balanced web servers. Had principal db in one zone, mirror in another zone, and witness in a 3rd zone. Had 3 web servers in each of those zones as well. Without warning, every 2 to 3 days the principal db would just stop serving requests, the requests would backlog and pile up, the database became completely unresponsive, auto-failover didn't happen and you couldn't initiate a manual failover either. Only way to failover was to shutdown the principal server. This happened numerous times with different instance types. Amazon couldn't resolve the issue but acknowledged that there was a bug with the virtualization hard drive driver that made their platform unfit for this purpose; the part that really pisses me off is that Amazon has public facing documents that state otherwise. (FYI: I run a low-use ms-sql express database in AWS - no mirror - and it works fine.)

I spent two months prepping to move to the nirvana of "the cloud". No more hardware to maintain, easy scaling, etc, etc. It really is a beautiful solution that holds so much promise. For now though, it was an epic effing failure. More downtime in 2 weeks with AWS than I had in the previous 4 years. Rolled back to my data center after 2 weeks with AWS and I still twitch when I think about the cloud.