Notes on PDC: Windows Azure
There were always going to be high expectations for Microsoft's 2008 Professional Developer Conference (PDC). This was the first PDC without Bill Gates at the helm, and let's also not forget that the PDC event scheduled for 2007 was unceremoniously cancelled at the last minute - fuelling speculation that Microsoft's product roadmap was in the process of being torn up. So, in 2008, the key question on everyone's lips was: would the event signal a Microsoft back on track with its developer story?
It turns out that 2008 has been a pretty solid year for Microsoft in terms of developer technology delivery, with the release of Windows Server 2008, Visual Studio 2008, SQLServer 2008, Hyper-V and Sliverlight v2. And at PDC 2008, in short, Microsoft did what it had to do. It provided insight into its strategy and an outline solution set for a cloud computing era; made good some of the mess that was the release of Windows Vista; and showed the world that even with Bill gone, there's still a strong management team with leadership vision and product foresight in place.
I'm going to tackle these points over a couple of blog entries. First, here, I'll tackle Microsoft's cloud computing strategy and the newly-announced "Windows Azure" initiative.
After a slightly slow start, with Windows Azure, Microsoft has now placed a strong bet on cloud computing and cloud-based applications. The company now believes that "the systems for cloud computing will be setting the stage for the next 50 years - with new patterns and new models of deployment, and application models for a world of parallel computing."
It was interesting to hear the company praise Amazon's innovation and exploration in this field with its EC2 offering. However, Microsoft is of the opinion that ultimately, it'll be in a better position to offer a more comprehensive end-to-end service portfolio than Amazon - owing to the fact that it can leverage strong pre-existing market positions with development tools, management solutions and server environments. As a side-note, of course, it's worth remembering that Amazon and the other "cloud innovators" counter this position by saying that new computing models don't have to rely on old tools and skills - indeed, disruptions can (and sometimes should) bring new tools and techniques that are most suited to the job in hand. When Microsoft was at the forefront of the shift towards client-server systems and away from mainframes, we don't remember it championing COBOL or CICS on the desktop.
Azure, Microsoft's cloud-based service solution, will be a hosting platform for applications and services that can be built by Microsoft, Independent Software Vendors (ISV), service providers and customers using a combination of Live, .NET, SQL, SharePoint and CRM services. Azure is designed to deliver services that can be leveraged rapidly and easily, and will be delivered by Microsoft's data centres in the US and across the rest of the world. In line with its positioning vs. Amazon, it will use its existing development tools and the .NET framework as the developer entry point to the Azure platform. What's also interesting, and encouraging, is that Azure is not just for customers: Microsoft is also aiming to use Azure to host its own internal systems.
Of course, Azure was expected - and widely trailed. In the coming months, we expect that most, if not all, the major software infrastructure vendors and service providers with sizeable data centres will launch some form of "cloud based" strategy for their product and services portfolios. The company's strategy to leverage existing technology and products wherever possible could be a good move - in that it could remove a potential barrier to adoption, and is likely to please many of its existing customers. However, given the side-note above, the fact that Microsoft is sticking to its existing development technology framework for Azure isn't a guarantee of market domination. In the immediate term, though, the challenge for Microsoft, as always these days, will be to ensure that it can articulate its strategy and product direction precisely and clearly. The breadth of Microsoft's portfolio and the number of markets that the company covers means that it's all too easy for the company to confuse its audiences with stories and strategies that aren't "joined up".
There was one important missed opportunity in the Ray Ozzie keynote which sketched out Azure, and it was an opportunity to explain the technical, regulatory and legislative demands that developers would likely have to meet in building application services for deployment on Azure - and to explain how Microsoft would help developers with the associated challenges. The issues was skated over very lightly, and this was something that a lot of people were expecting to hear about.
Many organisations are likely to struggle with implementing cloud-based services, and not only because of the technical challenges: there's an architecture and planning question to be addressed, too, which at the moment is not receiving as much attention as it might. The question is not about how to build services, so much, but *what* services to build, and *why*. This is a question that many organisations already struggle with in the context of SOA - which is one of the reasons why most SOA efforts today are still tightly constrained project-level efforts dealing largely with application integration use cases. For all these reasons, we expect the primary targets for platforms like Windows Azure - at least in the short term - to be ISVS and Service Providers rather than enterprise development shops.
Labels: Azure, Microsoft, PDC, Windows
