I picked this up from Brady, that S3 now has an SLA.
It's a good move.
However, there are risks associated with using a utility computing service, though there are risks associated with building a system in-house.
Tim Anderson noted that that the guarantee isn't worth much offering a fairly paltry maximum refund of 25%. That's not unusual for the hosted market, I've seen much worse. The real value in an SLA is to give an indication of what the provider expects the service to run at and a financial incentive to run at that level of service.
Unfortunately, though these are good moves - the greatest risks - that of catastrophic loss or a new guy in charge or a price hike or service closure still remain.
These risks will not be solved until there are alternative providers of the same service and a competitive utility market with freedom to move from one service provider to another (e.g. high levels of patration). For this we are going to need open standards and to avoid issues of loss of strategic control - open source implementations of such standards.
Of course this will led to undifferentiated price competition - something good for consumers but terrible for margins.
Well, I say terrible, but then again with the usefulness of such services being widespread, such a route allows for high levels of adoption and hence volume - as per other "utility services" like power.
If such a competitive market doesn't form, then as these services grow, I suspect as per Jesse's post we will eventually need government intervention.
So which standard will win out? OVF? (see Rich Miller's and William Vambenepe's excellent blogs for more on this subject).
Well it's early days, and the winners will emerge through wide adoption.
So it's a good move, but more is needed.