Research makes a stronger mark on Microsoft

MOUNTAIN VIEW, CALIFORNIA - Microsoft Corp.'s research group receives only a small piece of the company's massive research and development (R&D) budget, which in its current fiscal year is a sizeable $6.8 billion. Unlike what might be expected, Microsoft Research's (MSR's) main task is not researching new products. Instead, the group mostly does fundamental technology research.

Free IT resource Hear how top CIOs turn change into a competitive advantage. Sponsored by HP Free IT resource Attend the SOA Executive Forum: Breaking SOA Bottlenecks SOAExecForum.com/may2007 Sponsored by InfoWorld

Yet, the group, more than 700-people-strong, is more involved with Microsoft's products than before. Microsoft Research, for example, is working closely with Microsoft's product development teams on the development of Longhorn, the code name for the Windows release expected in 2006.

"We're actually a very small 'R' next to the very large 'D' that Microsoft does," said Daniel Ling, corporate vice president and head of Microsoft's Redmond, Washington, research lab in a presentation Wednesday. Yet, "every single Microsoft product has been touched by Microsoft Research," he said.

Ling spoke at a Microsoft Research road show at the software maker's Mountain View, California, campus. The group presented 10 of its projects, including future technologies designed to better secure computers, detect spam on Web pages and create summaries out of several news stories on a single topic. After his presentation, Ling sat down with IDG News Service to talk about Microsoft Research, the role the group plays within Microsoft and the increasing impact it has on Microsoft products.

IDGNS: You said that every product that Microsoft offers today was touched at one point by Microsoft Research. Is that something that you see increasing? Are Microsoft product groups using your services more than ever before?

Ling:I think over time it is true that we are expanding our influence. For example, I would say that around four years ago, we had relatively little engagement with MSN and now we have a lot of engagements with MSN related to spam and on search-related issues. Over the years, I think we are gradually increasing our involvement with different parts of Microsoft.

IDGNS: How does that happen? Do the product groups come to you and ask you to research technologies for a specific product or do Microsoft Research employees set out their own research directions?

DL: It is a mixture of both. Sometimes, we think a direction is important and we will pursue it on our own, then later the product groups will come to us. A good example of this is data mining in SQL (Server). We started our data mining efforts in Microsoft Research. SQL actually was not focused on the data warehouse, data mining or any of those things. We incubated, with the agreement of the SQL team, the data-mining process within MSR and in the next release (the team) said, "Oh, this is really interesting, let's include that as part of the main product offering." So it happens like that.

And sometimes (the product teams) come to us and say, "Gee, this is a really important area, can you help us do something about that?" Clearly, security is becoming a more and more important issue for Microsoft, and we have done more on security and bug- related things to meet that need.

IDGNS: Is your collaboration with the product groups something that has also increased over the years?

DL: I think we have a very good history of working with the product groups -- really since the beginning of Microsoft Research. So I am not sure (there) is a change in the way we work or a cultural change or anything like that. To give you an example, the original video and audio streaming work was actually a research project that started in the early to mid-'90s ... and later grew in to the whole (Microsoft) digital media division.

IDGNS: How long does it typically take for work by Microsoft Research to show up in Microsoft products?

DL: You should not think about these projects as product development. We work very closely with the product groups, so ideas from these research projects may appear in future versions of Microsoft products. For example, a lot of the spam technologies we developed showed up relatively quickly on MSN, by that I mean maybe six months or something like that. Longhorn obviously has a longer development cycle, so it will take longer to show up. That has to do partly with the technology and partly with our product cycles. It is not an easy thing to quantify.

IDGNS: If you look at your involvement with Microsoft products, is it mostly Web products such as MSN, consumer or desktop products such as the Windows client or enterprise products such as Windows Server, SQL Server or BizTalk Server? Is there any way you can see Microsoft Research being more involved with any specific part of Microsoft?

DL: I think we try to be involved with all of them. At events like today's, we tend to show the more visual things and not the back-end, tool kind of things. Take for example SQL. Here we have a very interesting project that tries to simplify the administration of a database and make it much easier to automatically tune the database to the workload that you see. This is a very enterprise kind of application. Our work cuts across all departments.

IDGNS: If I am a SQL Server user and tell my Microsoft representative that I need an easier way to manage my database, will this person tell me this is good feedback that can be used when the next version is being developed? You say that you are working on things related to SQL Server as well, so are customer requests coming down to you from the product team?

DL: No. We think it is important to give our researchers a lot of freedom in terms of choosing the topics that they want to investigate. That said, we ask the researchers to spend a lot of time, initially, with the product groups so they know what is going on and what is important. It is not that the product groups can say, "Definitely work on that." Very often we will work on something because we want to help the product teams, but they do not have direct control over the process.

IDGNS: So researchers don't go out to meet Microsoft customers?

DL: Sometimes they do. They might meet them on location or in executive briefings. There are lots of opportunities to interact with customers.

IDGNS: One of the research projects Microsoft talked about today is clustering and storage virtualization. There are companies already doing these things on Windows as well as Unix and Linux. What is Microsoft looking at developing itself?

DL: What we're trying to do is look at the next generation of clusters and build them in such a way that the amount of management required is minimized. We have a project called Boxwood. This is an innovative approach to building a new kind of storage infrastructure. (More details on Boxwood are at: http://research.microsoft.com/research/sv/Boxwood/)

IDGNS: The biggest thing that is happening at Microsoft right now is developing Longhorn and what Microsoft executives have called "the Longhorn wave of products." What is your involvement in Longhorn? And would you say you're involved more with Longhorn than you were with other version of Windows?

DL: I think it's been very good. When Longhorn was being formulated, there was a whole series of task forces around the company to look at what to do and how to do it. Microsoft Research was very heavily involved in these discussions. So, I think we have played a key role in defining what should be part of Longhorn. As the product is being developed, we've been engaged pretty tightly with the various Windows groups that are involved.

IDGNS: Do you remember the development of Windows XP, for example? How does Longhorn compare?

DL: Yes, I think we are more involved with Longhorn than we were with XP. With Longhorn, because it is such an important milestone for the company, we've been heavily engaged.