SQL Server 2005 is fantastic. SQL Server 2000 was wonderful. SQL Server 7 was OK. I hear SQL Server 2008 will be even better...
...but wait a minute?? Really, SQL Server 2000 does everything I need. So does Oracle versions 6, 7, 8, 9, and 10! So does PostgreSQL, so does MySQL. So what gives? Don't get me wrong. I've grown up in the Microsoft garden. I still have a VB for DOS development kit, and I have used VB 3, 4, 5, 6, and .NET 2003 and 2005. These are superb and superior products. The madness is accumulating with a new release every 2 years. Microsoft is now forced to offer a downgrade from Vista to XP option because folks are getting fed up.
One client of mine is busy upgrading their flagship product from SQL Server 2000 to SQL Server 2005. Why? Not because of any new features. Not because of a better price. Not because of really any reason at all, except for the fact their customers are asking for it by name. Because SQL Server (and Oracle) are highly visible in the public's radar, many trade journals speak volumes of marketing info about the new bells and whistles they contain. As developers, however, we all know that perhaps the management UI is better, and 64-bit it great, but basically the engine worked fine and met all of our needs several versions earlier. So my client is spending $500,000 and seven or eight months upgrading their core product--and all of the internal support tools that must work with it--to SQL 2005. Again why? To 'add value to the customer experience'...meaning that the salesperson can say "yes, it works with SQL 2005."
You see, in the real world, often non-technical or semi-technical people make the final purchasing decisions for enterprise software. They grasp for 'clues' of what should separate an inferior product from a superior product, and--well--version 2005 should be better than 2000, right? Therefore, a product based on 2005 is better than a product based on 2000! Right? In reality, somebody must pay for the $500,000 so that same customer is misleading themselves into requesting a higher price-point.
A better model would be to choose a database that is solid, but not in the customer's radar. This way, they would have no 'journal information' about the database and would not be pressuring you to spend $500,000 every other year to keep up with Microsoft. PostgreSQL, MySQL, Interbase, etc. are all robust databases with 64-bit support now. If your software is for an internal client (your own company), this whole dynamic shouldn't affect you, but if you make a commercial client-server product, there's no doubt you've experienced this already.
Mike J Berry www.RedRockResearch.com