The issue still hasn't been corrected as of this writing, but today's six-plus hour outage of Amazon's Simple Storage Service (S3) continues to cast doubts as to the suitability of not only itself, but cloud-based services in general, for commercial, mission-critical applications.

Chronicled to some extent on the Amazon Web Services (AWS) Health Dashboard, the issues, which AWS characterizes as internal communications problems between S3 components, have effected commercial users such as Twitter (as if Twitter doesn't have enough availability and performance problems on its own) and Huffington Post as well as scores of others for an almost painful period of time today. Certainly heads would roll if critical storage systems in internal corporate networks were down, unplanned, for such a length of time. While AWS bills its service as "beta" it also claims to have evolved from internal systems which Amazon uses itself to host its high-traffic storefront operation... which hasn't been down for six hours in recent memory, if ever.

There are certainly new and different issues posed by offering AWS services to third parties, but despite the beta billing, and despite the potential for unforeseen uses, the problems exhibited by AWS this past year are causing even proponents of cloud computing (and I am one of these) to question either the concept or the specific implementation offered by Amazon. Initially, the company appeared to be among those best suited for offering such a service, but at this stage it appears if they have done almost exactly what the thumbnail sketch of the service was outlined to do: throw open the doors to some of the spare processing capacity on the Amazon network, without safeguard, segregation, or supervision. While a fine idea from the standpoint of profiting from their hardware investment, this is a notion that AWS has taken pains to dispel, by offering Service Level Agreements (SLA) and assuring customers of access and capacity repeatedly.

Growing pains are inevitable in any new venture, but the one thing that anyone should be able to demand is that those in charge of such ventures learn from their mistakes and avoid repeating them; while in the technical detail of the matters AWS may be forging new ground with this failure, the fact that they have not yet successfully addressed the meta-aspects of "How can we keep running when we have a failure?" indicates they aren't meeting that requirement yet.