Gates explains Longhorn changes
Published: 31 Aug 2004 11:05 BST
What is really causing sort of the rewrite on Longhorn? Is it these sort of demands from the partners, is it the technological difficulties of the project, or is it even personnel and execution issues at Microsoft?
There's no rewrite going on here. WinFS, I'd be the first to say, is very ambitious. Nobody has ever brought together the world of documents, media and structured information in giving you one simple set of verbs that lets you richly find, move around and replicate those things.
Ever since about 15 months ago, when we moved Peter Spiro of our database group to take charge of WinFS, we've made very good progress. What happened here is, as we looked at the new things we wanted to add to WinFS, that would have only been consistent with an 07 schedule -- adding the tabular stuff and figuring out a server plan.
So we definitely were faced with a decision that Jim, Peter, Steve (Ballmer) and I were having a lot of dialogue over these last couple of weeks. What was the right thing? Was it to take Longhorn as a whole and get these super-cool additional WinFS features in, knowing that that would push the release out into '07, or was it to come up with a plan that was a bit more clever and really not give up much?
The plan we have does give up WinFS shipping with Longhorn. And so if you want my basic assessment here, the glass is three-quarters full.
The WinFS team, in terms of its progress and performance, is doing very, very good work, but it couldn't take the additional features and make an 06 schedule. That's professional engineering on its part, to stand up and say, "no, if you want us to have those features, we're an 07 deliverable".
It seems as if software is taking longer to bring to market. SP2 kind of grew in scope. Things like Yukon and Whidbey have taken more time. Has software just gotten more complicated to write? What, if anything, does Microsoft need to do as a company to reflect that reality?
Our scheduling and predictability on this project has been better than it was on OS 360 (the mainframe operating system created by IBM). So software has not gotten more complex. Software with this kind of scope of features and compatibility has always been complex. That's the business we're in.
The ongoing dialogue we always have with our customers, ISVs and OEMs (original equipment manufacturers) is always one of two kinds. One is (that) we have a date-driven release. Things that make that date get in. For MSN, most of what it does is date-driven releases. It does very regular releases because of the nature of its market, and it doesn't have the compatibility challenge.
With the operating system, customers want releases to be on the order of, I'd say, two to three years. Hardware exploitation, media, security and wireless expectations mean that you've got to have that release. But it's enough of a challenge, in terms of deployment, testing and other things, that you wouldn't want two major operating-system releases in less than (a) two-year period.
Previous
Did you find this article useful?
178 out of 320 people found this useful
Digg
Slashdot
Del.ici.ous
Stumble
Reddit
Full Talkback thread
4 comments
