Arik Hesseldahl

Recent Posts by Arik Hesseldahl

Seven Questions for Doug Hauger, Head of Microsoft's Azure Cloud Platform

I had always been a little confused about Microsoft’s Windows Azure cloud computing platform. Amazon Web Services I get. But had you asked me to tell you how it and Windows Azure are different, I would have been a little hard pressed to tell you.

I can tell you that Windows Azure is going to make the telematics systems in the next generation of Toyota cars smarter. And I also know that this unit of Microsoft has been in a state of management flux recently. Amitabh Srivastava, the Microsoft Distinguished Fellow, who in 2006 took over a project then known only as Red Dog that went on to become Azure, left the company in February.

It’s no secret that, like so many other companies, Microsoft has some big plans for cloud services. It recently disclosed that it plans to spend more than $8 billion in research and development funds on its cloud strategy.

On a recent visit to the Microsoft campus in Redmond, I got a chance to sit down with Doug Hauger, Microsoft’s general manager of Windows Azure. And my first question was really really basic.

NewEnterprise: Doug, there’s so much happening in the cloud computing space these days, and most of the time when people think of cloud services they think of Amazon Web Services. And if they mention Windows Azure, they think, well, that’s Microsoft’s answer to Amazon. But you describe Azure as more of a platform-as-a-service. Can you walk me through the differences?

Hauger: Windows Azure started about five years ago. At that point it started because the company, as with all service providers, was facing some challenges on providing large, scalable, manageable services, not just to consumers, but to businesses that could dynamically scale, and that we could innovate on quickly, and bring out new features. Originally it was meant to be a platform we would use internally for services that we would then deliver out to customers. We quickly realized that we should sell it to partners and customers, and allow them to build on it as a platform.

There are fundamental differences between infrastructure as a service and what we did as platform as a service. It’s different in key ways from, say, what Amazon does with EC2 and S3 or VMWare being implemented in a data center. Our starting point for the design was to see the data center as a unit. That means the networking structure, the load-balancers, the power management, and so on–rather than in infrastructure as a service, you start from an individual server and move up.

If you allocate a service into Windows Azure and say you want it available 100 percent of the time, we will allocate it across multiple upgrade domains and physical power domains in such a way so that if any individual rack goes down or if we’re upgrading the operating system, there’s no interruption in service. That’s just a fundamentally different starting point, with an individual server and moving up. And the way that we do that is we have built out an abstraction layer of APIs that let you write to a set of services, storage services, computer services, networking services, et cetera. As a developer you can write to the service, and give us your application, and it just gets provisioned through what we call a fabric controller, that controls the data center, and also across multiple data centers. That was a design point. That’s how we allow people to write services that can scale and won’t fail and will be available all the time.

The conversation about infrastructure as a service typically starts at cost savings. You go see a customer and they say they want to cut their IT budget and outsource their IT, and so they start there. Platform as a service you start at the cost savings, but very quickly you see 10, 20 or 30 percent cost savings. But the conversation quickly turns to the innovation life cycle that they can get out of the platform. It’s much faster than you can at infrastructure as a service.

The big point that everyone gets about the cloud is that they can use it to save money, but then they quickly start asking what more can they do with the cloud. Are you seeing the same thing?

Yes, exactly. In the enterprise, they’re starting to turn the crank on innovation. I talk to customers who are turning things around in six weeks or a month whereas before they would six months or a year. I actually just talked to a customer the other day, and they said their developers were spending 40 to 50 percent of their time managing services and they couldn’t use that time writing software which was their job. When they moved to a platform as a service, they didn’t have to worry about that anymore. We’re seeing this happening in the enterprise where people are doing this for internal development and on services they’re building for their customers.

One example, Daimler just did their new version of the smart car. They wanted a service so you can check the status of your car when its charging from your smart phone, locate it, et cetera. They turned it around in a couple of weeks on Azure and launched it at the same time as the car launched.

We’re also seeing small players compete at the enterprise level. There’s a small company called Margin Pro and they do mortgage analysis and risk assessment on mortgages. Basically it’s a couple of economists and developers. They wrote the software on Windows Azure, and now they have 70 banks around the world, tens of millions of dollars in revenue, and they are competing with some of the biggest financial services companies in the world because of this back-end infrastructure data center they can use to deliver their results to their customers.

But do you have customers who run standard apps on it too?

Many standard applications have some level of customization, and so we’re seeing a lot of hybrid applications, where customers are extending them into Azure. We have a case with Coca-Cola Enterprises which has a back-end order-processing app that they’ve extended into Azure. And what they wanted to do was get more reach and more agility for the front-end. So they built a secure connection between their data center and Windows Azure and then extended the application out to their partners and customers, essentially people like Domino’s Pizza who order Coca Cola products. We’re seeing a lot of these cases of existing applications being extended like that.

We’re also seeing companies using the high performance computing workload. One example is a company called Greenbutton, which has done a high performance scheduling and billing system on Azure. Another is Pixar, which has an application called RenderMan, which does rendering. Most large animation houses have their own clusters they do this rendering on. Pixar wanted to open up a market for smaller animation houses, little Pixars if you will. They’re working with Greenbutton to embed their technology into RenderMan. They can farm their rendering out to Azure and be billed on a usage basis. That’s a case where you have a large company and a smaller one working together and leveraging the power of the cloud to open up a whole new marketplace where they can be competitive. We call it the democratization of IT.

At what point is the customers’ thinking right now? Are they still at that point where they want to see how much money they can save by moving things that are on-premise to the cloud or are they past that by now?

I would say there’s three buckets of customers. I’ve been in this role for three years and the conversations have evolved in some interesting ways. Three years ago I was telling people they should be adopters and get on board with this platform early. They all said to come back and talk to them in five years. Then about two years ago, the majority of customers were in the first bucket, interested in wanting to save money but they weren’t interested in doing any new innovation. And then there were a few willing to innovate a bit by extending their applications into the cloud. Today I would say many, but not the majority yet, but a lot of them say they get the cloud, they get the cost savings, and now they want to drive the innovation life cycle faster. And there is a growing percentage who are willing to do something completely different and compete in a new way and build a brand new business. It’s been exciting to see that.

What’s been really exciting has been seeing mid-sized companies realizing they can use the cloud to give them an advantage to innovate faster and compete against really big companies. So that is sort of the landscape. Interestingly, I’ve been seeing a lot more adoption among the financial services companies than I had anticipated.

Aren’t they the ones who are supposed to be the most conservative when it comes to IT? I mean, they’re aggressive on performance, but obsessed with security and so skeptical of using the cloud because they don’t want to let their data leave their hands.

Exactly. But think about financial services. They’ve been in cloud computing forever, but it’s just been running on their own proprietary clouds. And so they are very good about understanding their application portfolio, and what can run in a public cloud, what has to stay in a private cloud, and how they can span those clouds. You can basically say you want to do risk assessment on portfolios, you anonymize the data, and you run it on the public cloud, you do all the analytics, you bring it back on-premise and then you deliver it to your customer. Having that kind of mentality in that industry allows them to move very quickly.

Also, manufacturing is moving and adopting the cloud faster than I would have guessed. And interestingly enough, government–not so much federal, because there’s so many certification requirements–but state and local governments are embracing the cloud because of the economic situation, and these are not just governments within the U.S. In Australia and Western Europe, we’re seeing governments adopting and building out applications so they can get services out to their citizens.

So what’s keeping you up at night? What makes you worry?

There’s a few things I think about. While we drive customers to a very fast innovation life cycle, we need to stay ahead of that innovation life cycle ourselves. We’ve done a pretty good job with that. One example, when we first released in beta a few years ago, we had .NET but we didn’t have PHP or Java. We got feedback immediately, almost on the first day, that customers wanted those and right away. And so we turned it around and added those within three months. Our ability to turn the crank pretty quickly is there. And that is something that in the software industry and specifically Microsoft, we have to make sure we make this turn toward service delivery, where we have to innovate quickly so you can deliver services. I think we’re doing a good job, but it’s something top of mind for me.

What are they asking for now? Is there something new the customers want that they don’t have?

They’re asking for continued investment in Java. We have it now, but making it a truly first class citizen, which is what we’re focused on delivering. We also need to keep our ear to the ground around things like application frameworks, extending the modeling capabilities in Visual Studio and things like that. It’s just a matter of thinking about the developer. We need to understand what they want, that’s what we’re here for.

Latest Video

View all videos »

Search »

Just as the atom bomb was the weapon that was supposed to render war obsolete, the Internet seems like capitalism’s ultimate feat of self-destructive genius, an economic doomsday device rendering it impossible for anyone to ever make a profit off anything again. It’s especially hopeless for those whose work is easily digitized and accessed free of charge.

— Author Tim Kreider on not getting paid for one’s work