Simple-talk is one of my favorite sites, and now they have what will be a series on parallelism in SQL Server. This first installment is light and airy.
A few months ago, I got into a bit of a tiff on a Postgres email group when I had the temerity to suggest that query level parallelism is not only a Good Thing, but the only way to maintain performance as we segue from ever faster clocks on single thread/core cpu's to multi-thread/core/processor machines. That group assembled (I don't recall anyone joining in my defense) asserted that engine level parallelism (doling out queries to threads) was enough.
I've been arguing for years that RDBMSs (not just SQL Server) will be better applications if they're designed to the multi-core/processor/SSD machine. After all, at least the multi-core part is now fait accompli, so why not? The beneficial side-effect is that BCNF schemas, with SSD as *primary* storage, are fully feasible. They are the minimal data (bytes, that is) needed to fulfill demand, and since they are "fully" normalized, DRI implements most if not all of the constraints on the data. That's been written about here, a bit.
For all the heat that MicroSoft gets, even from me on occasion, they do get databases. Good on them.
For completeness, here's DB2 docs, rather dry, but then...
And here's Oracle.
Neither is as seamless, at first blush, as SQL Server. There, I said something nice about a MicroSoft product.