Microsoft's Cloud Chief on Windows Azure, Tools and More
The approach you guys have taken has not been marketing or an attack mode. At the end of the day, code talks. And at the end of the day, code usually wins. At the end of the day, people are going to use the product that works best. For traditional enterprise vendors, the real danger of the cloud is as soon as you pay for what you use, an awful amount of enterprise software sits on the shelf. And suddenly the rules change in terms of licensing or usage and how you sell it. So having a product that's easy to use, that's feature rich, that meets customer expectations and is moving fast… if you're not doing that you're in big trouble. And it's not just about Azure, it's a whole portfolio of commercial products, whether its Windows Azure, whether it's Visual Studio Online, whether it's Office 365, we've made the transformation to a devices and services company. And we made that as fast, and probably deeper, than the transition we made in 1996 on the Internet. We added the browser to all of our products in 1996–sometimes superficial and sometimes deep. But if you look at the investment we made over the last two years in the cloud as a company, it's a much more major fundamental shift and a much bigger bet.I never worked with Ray on Azure. He left Microsoft before I joined Azure. I worked with him on client things and I had a good relationship with Ray, but I never actually talked to him around Azure. One of the things that I think Ray was very farsighted on is the desire that customers have around the power of more platform as a service. And having more managed services that you're not having to, as a developer, maintain, patch, tweak and operate and that just work. I think that is something that all cloud vendors are starting to kind of pursue. And at the time when he was evangelizing it–I think Azure first went public in early 2008–that was a very farsighted thing. I think one thing that the original Azure vision… I think if we could go back in time we would have introduced Infrastructure as a Service at the same time. I think one of the things we looked at was where is the market going to be six or seven years from now. And one of things that we've seen since we released our Infrastructure as a Service--I think Satya has been quoted as saying at least 20 percent of our compute is now IaaS–is both the desire of companies that want to be able to reuse their existing code or the flexibility that it gives you as an on-ramp. And in some ways I think our PaaS story has gotten stronger because we can say we now have Infrastructure as a Service as well. Sometimes people would say, "Oh, but you're only PaaS," and they'd claim there was lock-in or whatever. But when you can say, "No, you can run any random piece of software you want, any data store you want, if you want to run this new NoSQL instance on this version of Linux, great, we support that." Then people tend to say great, now tell me about your PaaS stuff. Now that I know I can do anything, tell me why your stuff is better. And if I look at some of the PaaS abstractions we've done over the last 12 months, whether it's Web Sites, or Mobile Services or Media Services, those are examples of very pointy PaaS abstractions that we've seen huge uptake with. But I think we're going to see similar with BizTalk services and we're seeing more demand than we can almost handle on Hadoop right now–where people like the fact that it's open and using the existing frameworks in those cases. They can run Node or PHP with Web Sites or with Engine Yard. So it's what they know but it's provided as a finished service so they don't have to manage as much.
What do you think about where we are today with Azure and how much it jibes with what Ray Ozzie and his team set out to do and his vision for the platform?