EC2 Troubles Must Be Taken in Context
Proponents of cloud computing aren’t going to like the fact that Amazon had issues that resulted in outages among its EC2 customers’ sites. The know-it-alls out there are probably already saying, "If Amazon has issues like this, imagine what would happen if you placed your bet on a less-experienced cloud vendor!" The gravitational shift toward the cloud for both core and non-core systems has surely slowed down.
But the fact is that most insurers have their own outages when they host applications internally, in some cases with more frequency and severity than we’re seeing here with Amazon. It's interesting to note how some of the customers who are known to be affected have reacted. "We wouldn't be where we are without EC2," said one. So despite the horror of having their public-facing site go flighty for a day (or two--we're hearing the problems are not completely resolved), there's apparently a reserve of goodwill that has built up over many months of near-flawless operation.
Instead of putting the industry on red alert, Celent believes this event should focus the discussion on the relative reliability of various approaches, and the tradeoffs between them. Should you know your vendor's architecture and reality-check their DR and failover strategies? Absolutely. You should also run the business case for change, especially if gaining scale quickly, moving nimbly into new markets, or handling seasonal spikes in activity are issues for which you have few answers. Outages caused by a vendor are never a good thing, but they are probably not your biggest, baddest problems either.