Barry Schwartz, a professor at Swarthmore College, wrote a fascinating book called "The Paradox of Choice: Why More Is Less." His thesis is that people fall into two basic groups: maximizers, who always try to make the best possible choice (and who are often unhappy with better alternatives that come along later), and satisficers, who generally make choices that are "good enough" and then live with them.

What does this have to do with Exchange Server? It's all about hardware. Both Intel and AMD have been shipping large quantities of 64- bit CPUs, and server manufacturers have been taking advantage of their availability to ship x64 capability even in fairly low-end systems. For example, Dell and HP both offer x64 servers for list prices less than $2000. It's a safe bet that when you buy a new server in the next year or so from a major vendor, it will be x64-capable. Today, however, that's not universally true, particularly because much of the action in the low-end server market is centered on inexpensive Celeron and Pentium 4 parts.

Windows Server 2003 x64 Edition and Microsoft SQL Server 2005 are the only Microsoft products that support x64. The Exchange team hasn't said publicly what its x64 plans are, but I think it's safe to assume that Exchange 12 will support x64 when it ships. The hugely expanded address space means that you can pack a lot more RAM into x64-based servers, which would be useful for the caching that the Exchange Information Store and DSAccess processes perform. The x64 extensions might also allow for other scale-up changes.

That's where the maximizer versus satisficer issue comes into play. If you're just now moving from Exchange Server 5.5 to Exchange Server 2003 (as many companies are), it might not make sense to spend extra money for x64 hardware when solid, inexpensive x32 hardware is still available. Satisficers would probably just grab the best hardware available now and happily use it, upgrading later when they're ready to move to Exchange 12. Maximizers, however, would probably prefer to wait until the last possible minute to buy hardware, because capacity and prices improve over time.

It turns out that this is a false dichotomy in some respects. Exchange 2003 doesn't, and won't, run on the x64 version of Windows 2003. So if you buy x64 hardware, you won't be able to use the x64 extensions right now. That's OK, though, because x64 capability doesn't always come at a premium, especially when you buy a server that includes Windows 2003. The x64 version doesn't cost anything extra. (Although I don't know what it costs to upgrade an existing Windows Server installation to x64 if you don't have Software Assurance--SA-- and this might weaken my argument.)

Many sites aren't looking at Exchange 12 for deployment yet; they're deciding what hardware to buy in the remainder of 2005 and in 2006, and they'll probably still have that hardware when they get ready to deploy Exchange 12. In that light, and given the low (or zero) cost differential, it makes sense to plan to buy x64 hardware now, then move to x64 Exchange when it's available and when it suits your business requirements.

Drop me a note telling me whether you currently have any x64 hardware, whether you're planning to upgrade your hardware between now and the end of next year, and why you would--or wouldn't--consider moving to x64 hardware as part of your upgrade cycle.