Yesterday was a bad blue monday for customers of Microsoft's hosted business software in North America. A network failure disconnected them from their business productivity suites and hosted mail servers for over two hourse. To make matters worse, it happened at the start of the working day, when people actually DO read their mail and start planning the rest of the week.
"The outage was caused by a network issue that is
now fully resolved, and service has returned to normal. During the
duration of the issue, customers were updated regularly via our normal
communication channels. We sincerely apologize to our customers for any
inconvenience this incident may have caused them," reads Microsoft's statement.
Resolved or not, outages like this one are simply unacceptable. Two hours of productivity is a lot, and more importantly: it COSTS a lot. Yes, it can happen with services like hotmail, MSN and the like, but this is corporate stuff we are talking about: BPOS, Exchange Online, Sharepoint Online and all that.
It also lays bare a problem that Cloud Computing is experiencing all over the board: doubts about availability. A problem on the network level immediatly has serious consequences for everybody. SLA or not, downtime is something to be avoided. In the 'traditional' model of computing, a problem could spread out the way this did, but usually it stays quite local. With the cloud, every network problem automatically means an availability problem.
And it is not as if this happened at a small start up vendor. The fact that it happens to Microsoft means that this can happen to everyone (even though users of hotmail might think otherwise). For the cloud, bad press exists, and incidents like this will shoo away potential customers back to their own data centers.