Amazon crash reveals 'cloud' computing actually based on data centers

Cloud requires the same care, skills and sophisticated management as other areas of IT.


Amazon's cloud crash is turning out to be the IT world's equivalent of the rubberneck-attracting, traffic-snarling dramatic highway disaster of the season.

There's a promising new technology that has been promising (or threatening, depending on your point of view) to shake up the structure of the whole industry.

It has a celebrity victim that built a bulletproof network of data centers to support its own business, then opened them up to the public.

It has innocent bystanders flattened by the impact, several of whom suffered serious and long-lasting injuries.

It also has an object lesson to anyone who thinks of cloud computing as a quick path to problem-free, low-cost, top-flight computing without the trouble of building or maintaining the technology to make that possible.

Here's a tip you can pass on to the business-unit managers who are either buying cloud services themselves, or can't understand why you can't get rid of your expensive data centers and just lease some from Amazon, like the shiny BMWs the sales staff pay rent on for three years, then give back without any equity to show for it:

Clouds are data centers. They're big data centers that are at least as complex as your internal data centers and it takes serious technology chops to build, distribute and run n-tier applications across their many facilities to get the kind of performance and reliability the business units think comes for free.

Apps running in the cloud are inherently multi-tier – housing data on one set of servers and SANs, applications on another set, databases on a third, networking, load balancing, resource-management, backup, disaster recovery, security and a dozen other functions on a third or a fourth or a fifth.

In large-scale clouds, those sets of servers aren't grids or clusters, they're data centers.
At Amazon one of those data centers -- the one housing the Relational Database Service that is to Amazon's cloud what an ordinary Oracle RDBMS is to a normal business application – stopped talking to the apps and APIs and message-queuing and other services that all relied on it.

That brought down, slowed down or stopped in their tracks sites that built their whole businesses on the assumption Amazon knew what it was doing when it built the cloud and when it said the cloud was a reliable platform on which to build.

That was true, actually. Amazon did.

Join us:






Answers - Powered by ITworld

Ask a Question