I follow up on a running totals post with an article describing different ways to achieve the same types of results for more complex grouping and aggregation.

I demonstrate how IDENT_CURRENT cannot be relied upon for reporting your new IDENTITY value, even under SERIALIZABLE.

I talk a little more about in-memory table-valued parameters, and explains why some scenarios you test may not demonstrate the power of this technology.

I explained how I justified purchasing a new Mac Pro, and show that it is very tough to order a comparable PC at similar price points.