I was recently invited to a Microsoft Developer’s Conference here in New York, and along with the muffins and the buffet lunch buffet was served a heapin’ helping of Azure, Redmond’s nascent cloud computing platform (currently in beta).

Despite the downside of potential privacy and network performance issues, cloud computing offers a lot of advantages (scalability, cost effectiveness, and ease of maintenance,  to name just a few).  This makes internet based, service-oriented computing a very attractive option (especially for small to mid-size businesses) - so we (along with almost everyone else) expect to see cloud computing continuing to gain traction.  In other words, more of your local CPU cycles are going to be moving from your desktop or local server (both probably Windows machines, I might add) up into the cloud.

Somebody’s cloud, that is - but whose?

Microsoft would prefer it to be theirs, thank you very much.  However, there are two primary competitors also in the marketplace: Google (with its Application Engine), and Amazon (with its EC2 “Elastic Computing” service):

• EC2 allows customers to rent a variable number of instances of virtual servers,  which the customer configures as needed and then installs applications on.  Originally limited to Unix and Solaris operating systems, Amazon now offers Windows Server and several flavors of Linux as well.  High marks go to Amazon for flexibility, but maintenance and overhead is as almost as high as if the servers were in a standard data center (albeit a really nice data center…)
• Google’s App Engine takes a different approach - in short, there’s less maintenance and overhead, but also less flexibility.  The service is currently limited to applications written in Python, which users administer via a web console - the underlying operating system(s) are protected and shielded from the user.   App Engine is currently in “Preview” mode (Google having evidently singlehandedly worn out the term “Beta”), so pricing is not yet known.  More importantly, it also remains to be seen whether Google will make other programming languages available besides Python.

The idea behind these two services was to leverage largely pre-existing server capacity, infrastructure, and expertise.  Unfortunately, Microsoft doesn’t happen to have a comparable worldwide network of internet-optimized server farms laying around unused, and they do like to think big out there in Redmond - so they are throwing the long ball on this one: at last week’s event, I learned about plans to build out 20 immense Azure data centers strategically located around the world (Microsoft is literally fork-lifting in shipping containers full of servers…)

Together, these data centers represent a $20 bil investment - which by coincidence, almost matches the $20.7 bil Microsoft holds in cash reserves - can you say “betting the farm”?   (If a less PR-challenged company was undertaking something this impressive over the next year, I think we’d be hearing a lot more about it…)

Azure Technically, what I like about Azure is that it’s more of a true single “cloud operating system” than either Google’s service (too opaque) or Amazon’s service (too fragmented).  With Azure, you’ll be able to run Microsoft’s managed code (such as ASP.net and C#), Microsoft’s native code (C++ ), and via .NET, you can also deploy Java and Ruby apps - or any combination of the above.  At the same time, the underlying system housekeeping (and most importantly, the overall failover, data storage, scalability, and load-balancing) are all Microsoft’s problems - so it would appear to be the best of both worlds.  However, I feel the real value-add of Azure has to do with these 20 planned data centers and with the effectiveness of the Azure “Fabric Controller” at managing them - if done well, it could be pretty spectacular.

Hence the title of this post:

The Most Important Person At Microsoft… To the extent computing continues to move from the desktop up to the cloud, Azure will be critical to Microsoft’s future - and since the Azure team is only about 150 people, that does narrow it down a bit (sidebar: according to anthropologist Robin Dunbar’s well-known research, 150 also tends to be the maximum size for effective human social groupings across a surprising variety of cultures).  But back to our “Most Important Person” award: is it Azure team leader Ray Ozzie?  Nope.  Is it either of his lieutenants Amitabh Srivastava or David Cutler?  Nu-unh.  Steve Ballmer?  No sir.

Is it Jerry Seinfeld?  Wrong again.

In my opinion, the most important person at Microsoft is Debra Chrapaty, in charge of the Azure data center infrastructure - because while Azure is currently being tested within just a single Redmond data center, how well Microsoft’s Fabric Controller will manage the Azure cloud as it expands to 20 geographically-diverse data centers is both the initiative’s largest differentiating factor and its largest unknown.

(By the way, Azure represents yet another step in the Privatization of the Internet - more on that here.)

(And here.)

(And here.)