My jaw hit the floor on this tech note from Microsoft: with SQL Server 2005 and a SAN, you can have multiple SQL servers hitting the same read-only reporting database.

This is really only a solution for reporting loads that require more work from the CPUs than they do from the drive arrays. If the report queries spend most of their time waiting on disk access, then this won’t offer much speed improvement.

However, as an example, I’ve worked with analysis functions that did math-intensive work on a relatively low number of records. We initially coded these formulas inside stored procedures and SQL functions, but the CPU load was too high relative to the amount of queries we’d need to run. When we scaled out to full production load, we would need more CPU power than was available from a single server. We ended up recoding the analysis work in standalone Delphi processing applications, adding processing queues, and so on, but it was a lot of management overhead.

With this new feature of SQL Server 2005, we’d have been able to keep the work inside SQL, make the applications faster, and reduce turnaround time for the reports.

Niiiice.