QandA: Microsoft Readies Team Foundation Server, Atlas Framework

Microsoft is on track to release its Visual Studio 2005 Team Foundation Server this month, adding the highly anticipated final component to the new Visual Studio 2005 Team System development platform it shipped in November. In advance of that milestone, Microsoft executives S. "Soma" Somasegar, vice president of the developer division, and Rick LaPlante, general manager of Visual Studio Team System, sat down to talk with CRN about Microsoft's overhauled developer-tools philosophy and about forthcoming projects like Atlas, Microsoft's AJAX (asynchronous JavaScript and XML) tools framework. Below are edited highlights from the discussion.

CRN: The first release of Visual Studio Team System offers distinct versions for four roles: project managers, software architects, software developers and testers. What's not there yet is tools to bring higher-level roles like business analysts into the process. What are the plans there?

LAPLANTE: Right now we have roles that I like to think of as one degree of separation from the developer, but those are clearly not the only roles in a development organization. There's administrators, there's business analysts ... when we talk about the vision of Team System, it's about how you continue to march out and get more and more people involved. There will be more SKUs. There isn't a roadmap for those right now, but I would say that within the next year we should have some formal roadmaps about what's coming next, with the next sets of releases.

CRN: There's been tremendous interest in Microsoft's plans for Atlas. How is that going to be deployed when it comes out?

SOMASEGAR: Today, it's going to be a plug-in to Visual Studio. We will have a GoLive of Atlas in the near future. Then it will be baked into the next version of Visual Studio, "Orcas," when we do that. Atlas is currently available as a CTP [community technology preview].

CRN: It's going to be standards-compliant? I know there have been concerns about Microsoft taking AJAX-like functionality in a proprietary direction.

SOMASEGAR: It will be very compliant. We do want to have a standards-based client, so that if standards are the only thing you care about, we have support from a framework as well as a tooling perspective. As you go up the stack to the higher end, where we have the Windows Presentation Foundation, then, hey, if you want a rich-client application, a smart-client application, then we have that support as well. But guess what, that's a Windows-specific thing.

LAPLANTE: We're looking at AJAX and saying, it's too hard. The programming model is too complex. Wouldn't it be good if you had a control model? Well, we have one. For ASP.Net.

SOMASEGAR: Atlas is a way for us to say, how can you do AJAX-style programming using a set of tools from us, particularly for developers who are used to that ASP.Net programming model. How can we bring some of those language constructs, some of those controls, for the client side that enable to you to do AJAX-style programming in a better way? That's the fundamental notion.

CRN: OK, so Orcas will include native Atlas support. What else will be in that release?

SOMASEGAR: Rick's team is finishing up Team Foundation Server, and the rest of the development team is planning what we want to do next. Hopefully over the next two months or so we'll be able to finish our planning and start making a roadmap.

LAPLANTE: When you ask what we're going to do next, it's the same thing we did the first time around. Six months of customer research. We want to get back out there and sit with the customers and figure out the pain points, starting with the strategy.

Let me give you an example. We as a business had the same problem. The Powers That Be stood up and said, 'We need to have better customer visibility. We need to have one way of talking to the customer.' It came from the senior levels of the business and got handed over to the IT organization and it touched no less than seven different IT organizations and 20-some systems. For the CIO, it's a very easy thing to say 'We need to do this.' But that translates into thousands of lines of code and million of dollars and scores of projects. Managing that matrix is what we're about. Historically, we've been about individual developer productivity. With Team System, we're about team productivity. I think we're going to be about organizational productivity. That's where we're going.

SOMASEGAR: We're trying to engage the community on an ongoing basis right from day one. There are big steps forward that we took with Visual Studio 2005. One is that we said we don't want to wait for beta releases anymore. We said, every four to six weeks let's take the build and share it with the community. Give us feedback, so that we have an opportunity to take that feedback and build it into the product.

CRN: That sounds like the open-source development model.CRN: That sounds like the open-source development model.

SOMASEGAR: It's all about community engagement. To me, that's the thing that I personally learned from the open-source world. The way they embrace the community -- it's a huge step that helps you build the right thing, and I'm a big fan of it.

For Orcas, we want to take the next step forward. Every specification we write for a feature in the product, we want to be able to share with the community and get their feedback in a timely manner. Day in and day out, I want to have a dialogue with the customers so that we can be building the right product. That's our vision.

CRN: With the Team System release, what's changed in how Microsoft approaches developer tools?

LAPLANTE: The core behind this is that we started seeing customers coming to us saying, 'We need to create value out of our IT organization.' Quickly, that came down to realizing how out of control engineering was. People were writing articles about 'taming the wild beast.' The natural conclusion to that is that people bring in the big-P word: process. 'We're going to fix this with process.' Which, in most cases when we talked to customers, exacerbated the problem.

When we started doing this, I was the guy doing all the research. For a couple of months I was depressed. I was like, 'It turns out it's not a software problem!' But it turns out that software can solve these problems -- it's just not the software we thought of traditionally. It's not the software for 'let me make you write a better test; let me make you write a better architecture.' Those are necessary pieces, but they're not sufficient. If they were sufficient we wouldn't still be so bad, and we are. People ask me, 'What do you think about the Rational software?' They had 20 years to solve this problem, that's what I think. And yet, here we are today. I think a new approach is clearly warranted. It turns out that approach is more about collaboration and visibility and people working together than it is about any particular tool.

We looked at all the tools that we have -- and we have a lot of state-of-the-art software inside of Microsoft -- and we said, there's nuggets inside each of those pieces but we're going to start from the beginning to build a system that does three things. First, increase the predictability of success. The second thing was, we said, we're going to do that without driving productivity into the ground and making everybody who is good want to quit. The third thing is what we call out-of-the-box productive Day One. That encapsulates many things, including an ease of use that just didn't exist in the market.

The anecdote that makes me feel we've done a good job is that the bulk of the sales and the pilot projects that we're seeing are initiated by the engineering teams. The excitement we've seen for the product without the server being [released to manufacturing] is actually pretty amazing. We expect the excitement when the server does go to RTM this month will be massive.