Windows Azure: Why Microsoft Better Get It Right

Microsoft unleashed a new cloud computing ecosystem at its recent Professional Developers Conference event, even though most observers chose to focus on more obvious, though less important, aspects of its announcement.

Essentially, Microsoft is going to create a Windows-based cloud infrastructure. Many of the details of its ultimate offering are still unclear, and it chose to discuss Azure primarily in terms of how it enables Microsoft-offered hosting (dubbed Azure Services Platform) of Microsoft applications like Exchange, Live Services, .NET Services, SQL Server, SharePoint, and Dynamics CRM, and SQL Server. I'm surprised that so many of the commentators (and presumably Microsoft itself) chose to discuss this offering in terms of Microsoft offering SaaS version of its products. While I think this offering approach is interesting, it falls short of revolutionary; by contrast, the revolutionary aspects of Azure have barely been touched on by all the commentary about the new offering. Let me offer my take on what's really interesting about Azure:
Microsoft offering hosted versions (aka SaaS) of its applications is interesting; however, plenty of people already offer hosted versions of its server products (e.g., Exchange, SharePoint). So the mere fact of these apps being available in "the cloud" is nothing new. However, Microsoft has some interesting flexibility here. Other businesses offering hosted versions of these apps have to obtain licenses for them and pay some amount related to the list price of the product. By contrast, Microsoft, as the producer of the products, can -- should it choose -- price its hosted version nearer the marginal license cost of an instance, i.e., near-zero. Of course, Microsoft still has to pay for the infrastructure, operations, etc., but it can clearly, should it choose, obtain a price advantage compated to competitive offerings. This leads us to the next point: infrastructure pricing.

Microsoft, based on the cash flow from its packaged software offerings, clearly has a capital cost advantage compared to its competitors for hosting Microsoft applications. And, based on its experience in hosting Hotmai, Microsoft clearly has operational experience capable of scaling an infrastructure cost-effectively. Added to its ability to price at the margin for software licenses, this obviously provides Microsoft with the ability to be the low-cost provider of Microsoft application hosting. And this advantage doesn't even include the (dare I say it) synergies available to it based on its common ownership of the cloud offering and the applications themselves.

However, focusing on these aspects of the offering is missing the forest for the trees, so to speak; perhaps a better way to say it is that it focuses on the low-hanging fruit without realizing there is much more-and sweeter-fruit available just slightly further beyond the low-hanging stuff. And that's where Azure gets interesting.

First and foremost, Azure offers a way for Microsoft-based applications to be deployed in the cloud. All of the cloud offerings thus far have been Linux-oriented and required Linux-oriented skills. This has been fine for the first generation of cloud developers: they're early adopters most likely to have advanced skills. There is an enormous base of mainstream developers with Windows-based skills, though; corporations are stuffed with Windows developers. Before Azure, these developers were blocked from developing cloud-based applications. With Azure, they can participate in the cloud-which is why other elements of the announcement relating to .NET and SQL Server are so important. These capabilities of Azure will accelerate cloud adoption by enterprises. So Azure's support of the Windows development infrastructure is a big deal.



Cloud Economics Meets Windows Infrastructure
But even that isn't as important as what else Azure will provide: the mix of cloud economics and innovation with Windows infrastructure. One can argue that other providers (e.g., Amazon Web Services) could offer the same capability of hosting Windows infrastructure capabilities, so Azure, on first blush might not seem so important. However, these offerings would face the same issue alluded to earlier, namely, Microsoft's competitive advantage available through marginal license pricing (of course, that might be subject to antitrust issues, so the advantage actually might be moot). However, and this is not in doubt, Microsoft undoubtedly has an unanswerable advantage in that it can extend its components' architectures-at least theoretically-to be better suited to cloud infrastructure. That is to say, Microsoft could, say, take .NET, which today is primarily focused on operating on a single server, and extend it to transparently operate on a farm of servers, scaling up and down depending upon load. And this is where Azure could-could-get revolutionary. Marrying today's widely distributed Microsoft skill base with a cloud-capable architecture based on established Microsoft component development approaches, APIs, etc., could unleash a wave of innovation at least as great as the innovation already seen in EC2 (see this previous blog posting for some insight about today's cloud innovation). In fact, given the relative skill distribution of Linux vs. Windows, one could expect the Azure wave to be even larger.

Of course, this is all future tense, and by no means certain. Microsoft has, in the past, announced many, many initiatives that ultimately fizzled out. More challenging, perhaps, is how a company with large, established revenue streams will nurture a new offering that might clash with those established streams. This is Clayton Christensen territory. It can be all-too-tempting to skew a new offering to "better integrate" with current successful products to the detriment of the newcomer.

Microsoft has a mixed track record in this regard. I won't make a prediction about how it will turn out, but it will be a real challenge. However, cloud computing is, to my mind, at least, too important to fail at.

Cloud computing is at least as important as the move to distributed processing. If you track what distributed processing has meant to business and society-a computing on every desk and in every home, etc., etc.-you begin to get an appreciation for why Microsoft has to successfully address the cloud. Azure is a bet-the-company initiative-and there's a reason they're called bet-the-company: they're too important to fail at. So Microsoft needs to get Azure right.
Bernard Golden is CEO of consulting firm HyperStratus, which specializes in virtualization, cloud computing and related issues. He is also the author of "Virtualization for Dummies," the best-selling book on virtualization to date.


Reference : http://www.pcworld.com/article/153216/ms_windows_azure.html?tk=rss_news