It's worth reading Nathan Marz's book Big Data[/url]. It's an easy read but as a MEAP there are a few glaring typos.
It goes into the Lamda architecture in depth and the though processes underpinning it.
The message is that it is not just the tech underpinning what Twitter do it is the manner in which they use it.
As the blog posts suggest, they are trying to solve old problems by coming up with new blends of existing solutions. What they are doing goes far beyond what SQL Server can do and yet the "One DB to rule them all" approach really is where SQL Server has been moving