Two years ago, we decided that it was time to start doing some serious deployments of 64-bit hardware in Microsoft's internal IT organization -- MSIT. We looked around for a scenario that would justify the cost of the 64-bit hardware. When we decide to deploy technology in MSIT, even though the software is Microsoft engineered, we require each deployment to meet a business need. This helps us ensure that we are not only building the right products for the enterprise space, but also that we spend our money and resources wisely.
After some discussion, we realized that an area where we could find good value would be to move some of our domain controllers to 64-bit hardware running the 64-bit OS. Our internal .dit file (the flat file that stores Active Directory data) was larger than 14 GIG. The 4 GIG memory limit on our 32-bit boxes made it so that domain controllers were spending valuable time using paging files during database access. So, after checking with the Directory Services developers, who agreed that the theory should work, we purchased a system and put it into production. Sure enough, we saw performance gains. MSIT engineers have promised to have more detailed information on the performance gains by this early next week. I'll update the site when I get the details.
Move forward two years. Now, we're in an engineering march to release the Windows 64-bit edition client and server next year. We've revisited our scenarios and are now in the process of upgrading print servers, file servers, and our internal Human Resources website to 64-bit.
If you have some feedback or questions about the 64-bit deployments, I'd love to hear them and respond.