Let's play a game. I engage in an imaginary one-sided conversation with a client, and you try to guess what the conversation is about. Ready?
"I've got this great new database solution you should consider for your mission-critical enterprise needs. It's much more manageable and feature-rich than your existing platform, significantly less expensive, and I have tons of evidence to prove it can meet your current and future growth. I know the technology is new to you, but trust me. This is the right way to go."
I bet you're thinking "that goofy Microsoft bigot Brian is trying to convince one of his clients that SQL Server on Windows 2000 is the right database for an important project." Ha! I tricked you! I was actually portraying "Brian the UNIX bigot" of 10 years ago trying to convince a customer that a UNIX-based database was the right solution for his corporate needs. A decade ago, most people had never heard of the Internet, client/server "wasn't quite there," and UNIX was still a four-letter word in the glass-house computing environment.
"Progress, far from consisting in change, depends on retentiveness. Those who cannot remember the past are condemned to repeat it"—wise words from George Santayana. Interestingly, the same UNIX folks who currently mock Microsoft's ability to effectively serve enterprise-class database needs find themselves in an historical retrospective—this time playing the part of establishment instead of revolutionaries. Perhaps they've forgotten the past.
Consider the following. SQL Server recently shattered the world record for the Transaction Processing Performance Council's (TPC's) TPC-C benchmark. In fairness, Oracle can rightly claim the world record for performance on a single SMP node, and you can easily argue that one SMP box is simpler to manage than a federated database of multiple SMP nodes. But Oracle now maintains it doesn't like "the way SQL Server scales," rather than "SQL Server can't scale." That message is a fundamental shift in position and market reality.
Today, SQL Server is posting world-record benchmarks by using 8-CPU Compaq servers running 550MHz Intel Pentium III Xeon chips. Within a few months, Microsoft will be able to showcase a single SMP node running Windows 2000 Datacenter that supports 64GB of memory, powered by 32 Intel-class CPUs clocked at 1GHz. Isn't this soon-to-be-reality system scalable? Isn't it massively powerful? Hasn't it closed most of the gap between UNIX high end and Windows high end? Isn't it possible that the revolution UNIX folks fanned 10 years ago has reignited under the banner of Windows technologies?
While researching this week's column, I ran across some additional words of wisdom from Mr. Santayana: "Skepticism, like chastity, should not be relinquished too readily." I'm not suggesting that we should blindly accept Microsoft's claim that it can be all things to all people. And I'm not suggesting that SQL Server is beyond a shadow of a doubt 100 percent scalable enough for your needs. We should always approach new technology from any vendor with a healthy dose of skepticism. But discounting SQL Server as an enterprise-class database simply because it wasn't up to snuff 2 years ago is closing your eyes to progress. With that kind of thinking, you might as well pull out your trusty abacus, total up the extra bucks you'll spend on a comparable UNIX solution, then saddle up your horse and ride on out of town.
P.S. Microsoft Tech-Net and the Professional Association for SQL Server (PASS) are cohosting a SQL Server 2000 Live Chat Wednesday, April 26. Microsoft plans to release the SQL Server 2000 public beta version later this month, and this is your chance to ask product managers and developers your technical questions about Microsoft's next database and analysis system. You can submit questions in advance to firstname.lastname@example.org and participate in the chat by pointing your browser to PASS's chat page at http://www.sqlpass.org/chat.cfm.