I mess with Microsoft's Manos, he messes with me...

"Somebody messes with me, I'm gonna mess with him," says Robert De Niro, playing gangster Al Capone in one of my favorite movies, The Untouchables.

That line popped into my head while reading Mike Manos' vigorous, occasionally-mulish rebuttal to my article last week that had, I have to admit, the rather provocative title, '6 reasons why Microsoft's container-based approach to data centers won't work' (and, yes, the even more goading kicker, 'Are you listening, Microsoft?')

It turns out Manos grew up in the south side of Chicago, where Capone was also from. The environment, he writes, made him "ornery" and "someone who enjoys a good tussle." Maybe the same movie quote popped into Manos' mind.

My second thought while reading his "Stirring Anthills" blog was: if all I have to do to get some real answers out of Microsoft is piss off their executives, then ... no problem!

Seriously, while Manos had been kind enough to respond twice while I was working the story, I still felt he omitted some key details. His blog, while exhaustive, supplied many of the answers I and others sought, though, again not all of them.

For instance, Manos claims that containers are up to 20% cheaper for Microsoft than conventional server deployments, but doesn't back it up. I'm not looking for Manos' spreadsheets, but a little more detail would help convince me.

He also responds to the charge that each container creates a point of vulnerability by saying that the input/outputs for power, water, networking into the container are well-protected from accidents.

"You would need a mighty big shoe size to trip over and unplug one of these containers," he says. "Bottom line is that connections this large require electricians for installation or uninstall." Again, just enough detail to tantalize, but not enough to fully convince.

In response to doubts about whether or not providing enough backup generators to power all of the servers is wise, Manos reiterated that physical redundancy and disaster recovery are only options, not necessities when your apps are hosted across thousands of servers in multiple data centers worldwide.

"Look, many industries like the financial and others have some very specific guidelines around redundancy and reliability," Manos writes. "This drives tens of millions to hundreds of millions of extra cost per facility. The cloud approach eliminates this requirement and draws it up to the application." If that's so, why was there still a widespread Hotmail and Live Messenger outage as recently as two months ago? And would Mr. Sarbanes and Mr. Oxley agree with Microsoft's 'look ma, no redundancy!' stance?

As skeptical as I may sound, I actually find containers totally fascinating. As my wife well knows, I've brought up the idea of building a prefab home out of containers more than once.

But there are still enough lingering doubts, some about Microsoft's answers, some about areas where Microsoft refuses to open its kimono, that I remain guarded about the long-term success of Microsoft's plan, and even more that this represents a true paradigm shift for data centers. And I'm not the only one, as SearchDataCenter.com found out.