W2K Ship Date is Less Important than Its quality - Product Development

ENT, Oct 20, 1999 by Scott Bekker

It's beginning to look like those nasty jokes about the Windows 2000 name change had some merit. A year ago, when Microsoft Corp. rechristened Windows NT 5.0 as Windows 2000, cynical industry types speculated that Microsoft was really changing the name so it could slip the release date into calendar year 2000. At that point, the new version of NT was scheduled for a 1999 release. With 72 days to go until the change in years, Microsoft seems unlikely to meet the 1999 target, and indications are that Redmond has abandoned the goal. At this point, it's the right decision.

This upgrade has been in the works for a long time. The Beta 1 program for Windows NT 5.0 began in September 1997. The ship date slipped and slipped again as Microsoft kept adding features to try to stay ahead of evolving user requirements.

Once the beta program began to meet some semblance of a schedule around the Beta 3 release in April, rumors of an October release were widespread. Those expectations sagged to a Release to Manufacturing (RTM) of late October, with the product shipping by Comdex on Nov. 14; then to an RTM by Comdex, with Windows 2000 hitting the shelves in early 2000. Then there was much tea leaf reading about some comments from Bill Gates that the code would be finalized by the end of the year. All through this roller-coaster ride the official Microsoft line on the operating system has been that the beta program will last until customers are satisfied. It's the right thing for Microsoft to be saying, and it's the right thing for Microsoft to be doing.

A recent decision on application load balancing may indicate that Microsoft is serious about addressing user concerns. Microsoft took a bruising on the announcement that it was removing COM load balancing from Windows 2000 Advanced Server. Load balancing was one of the most heavily promoted new features of COM , the overhaul of COM/DCOM for Windows 2000. But beta users reported problems with the reliability and management of the tool -- so out it went.

Microsoft could have released Windows 2000 with the feature as it was, complete with a single point of failure and without the manageability tools necessary for many customers to use it. Microsoft could have rushed a manageability rework, potentially inserting new bugs into the operating system that customers would have to live with. But to the company's credit, it didn't.

Few users will care a year from now whether Windows 2000 went to manufacturing in December 1999 or 60 days later in February 2000. According to IT market research firms, few users will be deploying Windows 2000 until later in the year anyway, and few will be testing the operating system rigorously until after they've gotten their Y2K aftermath out of the way. What users will care about is whether the operating system crashes regularly or contains careless security holes.

Hopefully Microsoft's apparent intent to not cut corners or compromise quality to meet an arbitrary internal shipping date will hold. This time around, Microsoft is aiming for a more demanding class of customers who expect vendors who promise rock-solid systems to deliver rock-solid systems.

Ultimately, the success or failure of Windows 2000 will turn on whether the product is reliable, scalable and available when users deploy it -- not what day it ships.