We have a tremendous amount of respect for Amazon. They’re a fully self-built company who not only blazed their own trail, but effectively built the concept of e-commerce in a time when most people were still hearing “e-anything” for the first time. Their cloud hosting platform is arguably the most advanced, most stable platform in existence today. It’s the go-to service when money is no object. It’s the very best of the best. Put simply, it’s the service we one day hope to be compared to.
So why did it go down?
Two nights ago, severe storms knocked out one of Amazon’s cloud datacenters in Ashburn, Virginia. While full details haven’t been revealed, it appears that the facility lost utility power, and their backup generators failed to start. And to be fair, this can happen to any provider — even us. But the beauty of cloud computing is that the entire system is supposed to be self-healing. When one piece fails — be it a single server, a full rack, or an entire datacenter — replacement components spin up and take over. Network traffic gets re-routed, server load gets re-balanced to alternate servers, and data gets pulled from an alternate SAN. The idea is that except for a few brief moments of downtime, the outage will fix itself before most users know anything went wrong.
It’s a beautiful theory, and when it works, it’s amazing.
But here, today, right this instant, it’s just not quite ready. Not for us, anyway.
It’s like when you’re making that first pot of coffee in the morning. You’ve got the beans ground and ready to rock, and the water is at a rolling boil. You really need that first mug, so it’s tempting to just take the water straight off the burner and into the French press. Yeah, the coffee might be a little bitter because you didn’t let the water cool to that magic 195- to 205-degree range first, but it’s still technically coffee. Good enough, right?
Cloud hosting technology is still in its relative infancy when compared to other technologies such as shared hosting or virtualization. Heavyweights like OnApp, Parallels, and VMWare are currently duking it out for market share, while the industry still hasn’t even settled on what, exactly, the word “cloud” means. We’re all pretty sure it involves lots of important-sounding industry catchphrases like “high availability” and “infinite scalability”, but the offerings and performance — to say nothing of price — from one cloud vendor to the next still differ wildly.
We’re currently evaluating several cloud platforms for their stability, ease of use, and cost effectiveness. In a few weeks, our key employees will be attending a major web hosting conference in Boston, where we’ll get to chat directly with some of the senior employees at each vendor. We’ll get to ask them some of our hard questions, and see how they’ll handle some of our scenarios. And once we’re satisfied that a vendor can deliver a platform on which we can deploy a truly remarkable cloud environment, we’ll begin deploying the server, storage, and networking infrastructure required.
(Incidentally, if you’re a vendor pushing your cloud hosting platform and you think you can outshine the Big Three, drop us a line.)
Make no mistake about it: Cloud hosting is the Next Big Thing. As the price comes down, it WILL replace shared and VPS hosting. And rather than jump on the bandwagon of whatever this weeks’ most popular platform is, we’re doing our homework. Whatever platform we settle on will be something we’ll be using for many years to come, so it’s imperative that it rock.
As it stands today, there just isn’t a single vendor that provides that perfect combination of everything we need — at least not at a price that makes sense for small web hosts like us. To that end, we’re holding off on launching cloud hosting until a platform that meets our criteria comes along.