May 19, 2011 at 3:24 PM
A cloud outage never fails to bring out the cloud commentariat (just like I'm doing in this space today!). A failure like the recent Amazon lights-out is followed by a round of written warnings about how the technology isn't ready yet, operators don't really know what they're doing, customers are clueless, etc. Just yesterday came a reminder that platform-as-a-service is still immature while platform stability remains a worry.
Most recently, Microsoft's BPOS (Business Productivity Online Services) went dark, leaving users without e-mail and other functions, for two days. Also, according to some reports, users received no notifications from the provider about what was going on or what to do.
Microsoft failures are always like office birthday parties for technology writers: a brief and happy distraction from the drudgery of many tech beats. And so the BPOS outage brought us stories about how Microsoft's cloud future is bleak, how you can't trust the cloud, and that when the world ends on May 21, the cloud will be the first thing to die.
Now, skepticism about the cloud is naturally a good and necessary thing. Skepticism in general is a key asset for an IT manager or anyone who signs the invoices for IT services. But that skepticism should lead to one thing: An honest sit-down between provider and customer.
It becomes clearer that some outsourcing customers and other businesses are adopting the cloud but are not always being well-informed about what they're getting into. Apparently some people are not asking the right questions. A recent report from IDC analysts in Brazil indicated that many cloud clients are surprised at the cost of the services they're using. The bill arrives, and they are taken aback.
"In some cases, business components such as service levels and responsiveness are not described correctly in the contract," Celia Sarauza, research manager for IDC Brazil, told Nearshore Americas. "It is impossible to know how much some service or application costs today because it is considered in a big contract with other services."
Certainly a cloud-service contract is a complicated thing, as is cloud infrastructure, but you have to wonder what questions are being asked by customers prior to moving up. You also have to wonder if cloud providers are dishing up all the details. It's their responsibility to be forthcoming — this isn't horse-trading — but it's the client's to be skeptical and dubious and interrogatory to the point of being annoying, if that's what it takes to know what you're paying, what you're getting, and what happens when things go wrong.
Now, all the talk in the world won't improve shoddy engineering. As Alex Camino wrote on this site: "Regardless of strategy, cloud or non-cloud, industrialization, as the next step in the evolution of the IT services industry, should start with a rigorous process discipline."
And that brings up a whole set of talking points for the cloud client. Say to your provider point-blank: "Tell me about your processes, and explain how they are rigorous." The conversation between provider and customer has got to cover the serious questions about the provider's processes, especially its processes for redundancy and for responding to disaster.
When Amazon's service failed recently, some businesses kept on running because their applications fell over to backup instances. You can bet that those businesses had had specific discussions — internally and with Amazon — about how to make sure their operation did not fail.
That might be the most obvious statement you've read today, but when you hear that some cloud customers don't really understand their contracts, or haven't insisted on adequate measures to keep their cloud-based functions running, or didn't ask how things work, or are surprised when something bad happens — doesn't that make you wonder?