RSSMore musings on cloud reliability
Predictably, yesterday's extended Amazon Simple Storage Service (S3) outage has elicited a flurry of Monday morning responses from bloggers and media. The reactions range from the incensed to the accommodating to the practical. I have to admit that my own take, yesterday, was perhaps more overblown than I would have preferred it to be as I survey it in the cold, clear light of morning.
A surprising number of pundits are dragging out the "don't put your eggs all in one basket" admonishment, including Phil Wainewright and Dave Winer. This is a bit disappointing as it seems to me at least to almost entirely miss the point of cloud computing; although at this point it may be good advice for anyone relying on Amazon in particular… but perhaps that just points to their failure to fully realize the cloud model in the first place.
Stephen Arnold has a more interesting theory: that Amazon's engineering practices simply aren't up to speed (on the other hand, as hank williams points out, the company's Elastic Computing Cloud service, EC2, hasn't suffered from the same sorts of chronic problems, despite presumably having been developed under the same auspices). This is more in line with my own appreciation of the situation. I don't think these periodic failures are endemic to cloud computing as a whole; indeed, I think it should be realized as a solution which is much less susceptible to such outages and requiring fewer user-level redundancies than more conventional hosting practices. This is contrary to conventional wisdom, but I think conventional wisdom has lost its head on this one. Businesses have long since figured out how to solve single-point-of-failure problems in their on internal systems. It's not easy or cheap, but it's possible, and what you are buying when you buy cloud services should be just that solution. The entire point of the model is to defray the costs of doing so to a point where businesses otherwise unable to afford such power and redundancy can cost-effectively share the benefits of such engineering.
On the other hand, perhaps Amazon isn't truly offering a cloud service in S3, but simply cheap storage. This falls more in line with the performance of the system-it fails about as often as you would expect some cut-rate, on-the-cheap offering to do-and with the expectations that many customers seem to have of it, which is that it be far less costly than other storage services. So perhaps we're not talking about cloud computing at all here but have simply gotten S3 mixed up with other Amazon services which more clearly are cloud-like such as EC2.
