The dust has not yet settled on the Amazon outage. Apparently service is still degraded for some customers (Network World). Meanwhile, of course, most pundits are writing about how this will slow cloud adoption. My estimation is it will do very little to slow down adoption-primarily because for small companies on-site hosting is no alternative-they could not create a redundant, elastic datacenter on their own. For these small companies, the likely outcome of this outage is not re-thinking IF they will use the cloud, but rather HOW they use it.While some think the elasticity and redundancy are inherent aspects of hosting in the cloud, it turns out neither are. In terms of this outage, it is the redundancy that is under the spotlight (regarding redundancy). My guess is that those Amazon customers suffering outages due to the EC2 outage were aware of this, and I would suggest they were likely gambling that such a wide outage would not occur inorder to save time, effort, and ultimately money. Architecting to take advantage of these capabilities and then paying to deploy redundantly are costs a small company may think they cannot afford.
One thing is definite, cloud computing on the scale of Amazon's EC2 will make for much more dramatic IT news. When you get that many eggs in one basket and the basket drops, it is indeed news worthy.
It has been estimated that 80 percent of network downtime is due to unplanned change. Usually when systems go down or networks fail, it is due to unknown, undocumented or unauthorized changes. When I see events like the one currently playing out at Amazon, it makes me think about the importance of having mature change control processes. Change Management Software helps organizations understand the overall business risks associated with introducing change into a stable environment and ensures the right stakeholders are involved in the decision.
You should not have to find out the hard way that this is important like Amazon likely did.