Into the guts of SQL Server

I've just spent a week doing the SQL Server internals course developed by the 8 leading SS MVPs who *don't* work for Microsoft. It was a really well constructed course, that included my favourite pass-times like "when MS says don't look at X, grab a hex dump anyway and decipher it like this ...". Having done the hard-core internals work on other databases was a useful baseline that showed me where SQL Server is in comparison with, say, Oracle and DB2. In short, it has a few bits of really cool technology that Microsoft seem hopelessly unable to communicate to the tech community, and a whole bunch of stuff that they think is cool and l33t, but in actual fact is 10-year old stuff that their competition mastered ages ago, and is now the bread and butter of undergrad comp sci degrees that have a "build your own database" course. When someone blindly defends cache-hit ratios as an unquestionably good tuning instrument, you know the kool-aid is mighty strong :-)

I'm going to run a bunch of the tricks I learned over the guts of SQL Server 2005, to see what they've improved or changed. Given I've already developed a point-by-point run down on their new "headline" features, and how badly it appears they're playing catch-up (and failing!), it'll be interesting to see if the engine under the hood has fared any better in the six years it's taken to get to market. I'm willing to be surprised, but somehow think I won't be.