|
Judge Jackson ruled that Microsoft stifled innovation by “plac[ing] an oppressive thumb on the scale of competitive fortune.” But business practices aside, no one can plausibly argue that the company hasn't shown flashes of technological innovation in its own right.
We need no better evidence than SQL Server 7.0 (code-named Sphinx), which had massive influence on IT purchasing patterns—and how vendors responded to them— in the late 1990s. As we all know, by introducing integrated data warehousing extensions in release 7.0, Microsoft literally made online analytic processing a mass market overnight (The OLAP Report estimates more than 600,000 seats at this point) and inspired a trend toward packaged analytic applications. It also made enterprise customers appreciate ease of use for the first time, forcing competitors to move that checkbox further up their priority list.
For that reason, I was surprised by the contrition with which the Shiloh (forthcoming as SQL Server 2000 this summer) development team addressed assembled journalists and reviewers at a workshop in April. Specifically, within the first few presentations it became clear that, in the development team's opinion, SQL Server 7.0—albeit very popular—is not everything it could have been. Each presenter touched on a theme more in keeping with Clinton-esque empathy than Gates-ian swagger: We've failed you in the past, but now we're recommitted to doing the “right thing.” What, Microsoft admit mistakes? Has the world turned upside down?
Based on that premise, Shiloh is something of a make-good—a refinement of the Sphinx architecture as well as an answer to the fact that the Internet matured in the midst of the previous release's development life cycle. Indeed, if that release was a declaration of independence from SQL Server's legacy “good enough” approach, as product unit manager Peter Spiro described the jettisoned Sybase architecture (Microsoft rewrote 75 percent of the code base between releases 6.5 and 7.0), then you could say that Shiloh represents the fruits of reflection after a difficult break from the past.
Improved reliability, availability, and scalability are the goals of this release; according to the development team, SQL Server's performance is “no longer an issue.” One main objective is to decrease the number of “knobs” you need to tweak to keep your application humming. (Some of the DBAs in attendance were fairly salivating over the prospect of online index reorgs.) The team insisted that Shiloh will support failover much more effectively, obviating massive workarounds associated with 7.0's Wolfpack technology. And Shiloh's support for multi-node partitioned views is Microsoft's first “down payment,” as VP Paul Flessner called it, on the full-blown shared-nothing architecture we've been promised in the next release (code-named Yukon).
For Intelligent Enterprise readers, however, the most interesting improvements may be in SQL Server Analysis Services (formerly OLAP Services). Scalability is the most obvious enhancement; in Shiloh, dimensions can contain as many as 10 million members in multidimensional mode or hundreds of millions of members in relational mode. This degree of scalability—a prerequisite for clickstream analysis—should compete with that of standalone OLAP products, and of course, those products cost a lot more money. Multi-server support, improved dimension management, linked virtual cubes, and English-language query support should also turn some heads. And Shiloh's new data mining extensions may accomplish what many vendors have sought for years: to bring data mining into the business intelligence mainstream.
Don't Forget to Vote
I also want to remind you that the ballot for the Intelligent Enterprise Readers' Choice Awards 2000 is now live at IntelligentEnterprise.com. If you have strong feelings about a certain product, now is the time to make them known in a way that matters: by casting your vote. Ain't democracy great?
|
|
|
|
|