Past Was Guide for Changes to Windows Development
About a year ago on its Redmond, Washington, campus, a member of Microsoft's Windows Vista team met with a group of journalists to face some tough questions about the OS.
At the time, it was clear Vista was not going to be the great success Microsoft had predicted, as many of the company's critical business customers were beginning to reveal they would wait for the next release of Microsoft's client OS instead of upgrading corporate desktops to Vista.
Among questions posed to Microsoft that day were how the company could have gotten Vista so wrong after five-plus years of development, and how much longer Microsoft could justify putting out major software releases that needed substantial bug fixes before they were fit for enterprise deployment.
It has long been the mantra among IT professionals not to "go out and buy that first release" of Windows, but "wait until the service pack comes out because there are so many bugs and issues," said longtime Microsoft partner and customer Scott Noles, director of technology and education at Microsoft customer Kinex Medical, a medical rehabilitation center in Waukesha, Wisconsin.
The same rang true for Vista, and even for its predecessor Windows XP, a solid OS still in wide use that nonetheless also required a major service-pack release to deal with critical security issues that plagued enterprise users.
Microsoft had no answers that day to the questions it faced about Vista. Fast-forward a year later to now, however, and the company does.
For the past several months, Microsoft has engaged in an extended public mea culpa about Vista, and in the past two weeks alone has given a series of press interviews to explain how it changed the development process of Windows 7, the forthcoming client release, to learn from the mistakes it made in the past.
"We know we're still learning, but we always want to make tomorrow better than yesterday was," Mike Nash, corporate vice president of Windows product management, said about the development of Windows 7 in a recent interview.
He said Microsoft's move in March 2006 to put former head of Office development Steven Sinofsky in charge of Windows development was a key driver of changes in the process. Sinofsky is now senior vice president for the Windows and Windows Live Engineering Group, and Nash credits him for bringing order to the group.
Vista failed among business customers for a few key reasons. One was that its premium hardware requirements made it incompatible with PCs that companies already had running in their IT environments. That meant that upgrading to Vista meant that businesses also had to update hardware, a more expensive proposition than recycling existing machines.
Another was that Microsoft peripheral and software partners were not fully prepared for the release, which means many third-party products on which business users rely didn't work with Vista out of the gate.
Gavriella Schuster, a senior director of Windows product management, cited the "stop-and-start nature" of Vista's development process as contributing to partners' lack of preparedness for the final release. Microsoft stopped Vista's development in the middle of the process to overhaul the security of the OS, a move that delayed its final release.
Microsoft has changed how it built Windows 7 in a few ways to learn from errors of the past, it says. First, the company decided to "define a feature set early on" and only share that feature set with partners and customers when the company is confident that they will go into the final OS, Nash said.
This should avoid any confusion among Microsoft customers and partners as to what new features Windows 7 will have, and also will give Microsoft a chance to stabilize those features as much as possible.
"We made the ecosystem's job really hard [with Vista]," Nash acknowledged, because so much changed in Vista between the beta release and its release candidate, and then again between the release candidate and the final release. Making Windows 7's release milestones more predictable should ease the preparation process.
The beta of Windows 7 that is now available is already feature-complete, although its final release to business customers isn't expected until November, according to Microsoft's timeline. Microsoft also is forgoing the release of a second beta in favor of one more test version -- a release candidate that will appear soon before Windows 7's final release.
To solve the overall PC-compatibility issue, Microsoft has said that all versions of Windows 7 will run on even low-cost netbooks, which often don't have the same amount of RAM or hardware power that full-featured PCs do.
IT professionals are applauding these changes across the board, from Sinofsky's influence on the development process to the existing Windows 7 beta Microsoft's produced.
"By bringing a leader from the Office team, Microsoft got someone who knew how to ship and understood the responsibility of revving a ubiquitous and infrastructural product," said Andrew Brust, chief, new technology at IT consulting firm twentysixNew York, who is already successfully using Windows 7 on a low-cost PC.
He said that while it's necessary to freshen products like Windows and Office that so many businesses rely on with new features, it's also necessary to ensure updating to a new version won't be too disruptive to the user base. In Brust's opinion, Sinofsky proved in his work on the Office team that he could balance these needs, and was a good candidate to revise the Windows development process.
Stephen Hultquist, a principal with Infinite Summit in Boulder, Colorado, also agreed with the changes Microsoft made to the development process. "I think they're right on" with Windows 7, he said.
Hultquist, who cut his teeth as an enterprise CIO and first worked with Windows when it was still a joint Microsoft/IBM project, now calls himself a "revolving CIO" for a number of small and medium-sized companies. He moved very few of them to Vista, opting to wait for Windows 7, which he is testing.
There's certainly no love lost between Hultquist and Microsoft software. He said he thinks Microsoft put out Vista before it was ready -- even after five years of development -- "because they felt their reputation was at stake if they didn't release something."
However, Hultquist is glad that Microsoft has changed the development process for Windows 7, although he still thinks the forthcoming OS "is what Vista should have been."
Others defended Microsoft, saying the company -- which designs an OS that must work with an incredibly wide range of third-party products -- can't be entirely blamed for the software and hardware compatibilities that plagued Vista and affected its adoption among business customers.
Arlin Sorensen, CEO and president of Heartland Technology Solutions, said Microsoft's business partners and customers bear responsibility for giving themselves plenty of time to test a new Windows OS before it's released to ensure compatibility with the existing IT environment. Heartland, based in Harlan, Iowa, is a Microsoft partner and Windows Vista beta tester and early adopter.
"Contrary to popular desire, stuff just doesn't work without some effort, and it does require planning and budget to stay current and competitive with technology," Sorensen said in an e-mail. "We pride ourselves in engaging with Microsoft early and often, and it has been a great asset to us and our customer base. ...That is a choice we make, and by making it we don't run into the kinds of experiences many just want to point fingers and blame Microsoft for."
Still, Hultquist noted that if you compare the first release of a new Windows OS to the first release of one of Apple's Mac OSes or an update to Linux, the "stability and reliability and compatibility of Windows -- all of those features that matter the most for an OS -- is not even close, not even in the same category" as the other OSes, which tend to be more polished.
Microsoft doesn't have the luxury of developing its hardware and software in a vacuum, as Apple does, nor does it have the benefit of an entire community of open-source developers who can tweak Linux to their liking. Still, Microsoft's need to work with a broader scope of partners and customers to ensure Vista can interact with third-party products doesn't excuse it from being such a poor release in the first place, Hultquist said.
"From an end-user's perspective, who cares?" he said. "It's really about value delivered to the end-user."
If a company can't deliver value in a finished product, Hultquist said, "I can't use it then, I have to wait until it's stable and it helps my business."
Microsoft is hoping that because of the changes in its development process, Windows 7 will be far more stable and lack the myriad complications of Vista, even before its final release.
With Windows 7, "our goal is to make sure customers can treat the beta of one of our products like a release candidate and treat the release candidate as a final version," Nash said. With Windows 7's release date drawing near, customers will soon be able to find out if Microsoft delivers on that promise.