ITIL vs The Cloud: You can do both … if you’re very, very careful

Like Tweet Pin it Share Share Email

Qualifiers are for lawyers, not writers.

“In the absence of exogenous factors it’s generally true that 2 + 2 = 4. Notwithstanding the above, accepted mathematics applies except when unforeseen circumstances lead to different conclusions,” is, to be charitable, ungraceful.

When accuracy and good writing collide, though, accuracy should win. It didn’t in last week’s column on ITIL and The Cloud. I erred on the side of good writing, leaving out a necessary qualifier.

The column stated that maintaining multiple versions would “… blow up the Cloud’s economic model, which is built on the assumption of commoditized infrastructures.”

It should have said, maintaining multiple versions would blow up many Cloud providers’ economic models.

Blanket statements are usually wrong and always suspect. Long-time subscriber Mark Eisenberg, a member of Microsoft’s Azure team, let me know that, in his words, “We do support each version of the roles back to 1.0. The required image is specified in the service configuration. Customers can opt for automatic upgrades or select manual in which case the new version is not utilized until the customer completes QA.”

Azure, a Platform as a Service (PaaS) offering, also provides tools to build manageability into applications equivalent to the tools you’d have available were you to use .Net for locally hosted applications.

Conclusion: Some PaaS alternatives allow you to continue professional IT management practices, should you choose to do so.

The qualifier (“should you choose to do so”) is essential, and points to two concerns raised by the correspondence and comments I’ve received over the past two weeks.

Concern #1: Mistaking ITIL.

It’s easy to get ITIL wrong because it isn’t a thing, it’s a collection. Among its parts is a catalog of the practices necessary for successful IT operations. Whether you buy into the whole ITIL package, the catalog itself isn’t controversial.

Unless, that is, you think IT shouldn’t be in the business of managing the availability, performance, and capacity of the technology in its care, or think it makes sense to put application changes into production without proper testing and a back-out plan.

So forget I ever mentioned ITIL. The question was and is whether Cloud computing is compatible with basic standards of professional IT operations.

The information provided by Mr. Eisenberg suggests that with careful vendor and product due diligence, it’s possible. As the Hotmail situation mentioned last week made clear, though, some vendors, for some products, push changes into production for all customers, like it or not. The Hotmail problem isn’t unique. Another SaaS vendor I know of put three releases into production in one year that changed its APIs without notifying its customers.

No regression testing. No back-out. Just sudden failures.

Some Cloud vendors will support your quest for professional IT operations management. Many won’t. And it’s astonishingly hard to find the answer to such a simple question as whether upgrades are optional or forced.

Concern #2: The parts aren’t the whole

Imagine a manufacturer with something to sell. Its staff carefully sources every component and verifies that each one adheres to the highest standards. It assembles them with duct tape and chewing gum. Not only is its CEO astonished when his customers complain about the shoddy merchandise he’s selling them, he rejects their complaints as absurd.

Much of the commentary in this series expressed confidence that Cloud vendors would solve the problem of substandard systems management without much difficulty.

They won’t because they can’t, any more than our fictional manufacturer’s suppliers could guarantee the quality of his merchandise.

Take a hard look at your current systems portfolio. Mid-sized companies commonly find themselves with a hundred or more applications, connected by a spiderweb of interfaces and running on a platform layer consisting of another hundred or more different products supplied by dozens of different vendors.

If The Cloud succeeds, it will replace some but not all of these products and vendors, and the “it” that will replace them won’t be a single vendor’s suite of products.

Even companies that dive headlong into The Cloud will find they’re still managing a complex multi-vendor computing environment, with an equivalently complex interface spiderweb.

No matter how well each Cloud vendor addresses its responsibilities, that’s only a necessary condition for creating a manageable environment, not a sufficient one.

What’s required is what’s required right now: Hard work and occasional brilliance on the part of teams of highly qualified technical professionals.

If you think otherwise it’s time to remember the ancient cliché that keeps you from being conned: If it seems too good to be true, it probably is.

Comments (1)

Comments are closed.