What I Learned Today: Amdahl's Law

Reading through Chapter Fourteen of Beautiful Code, "How Elegant Code Evolves with Hardware: The Case of Guassian Elimination," I came across a new term I should know: Amdahl's Law.
Gene Amdahl observed that the limiting factor to any attempt at parallelizing an operation or series of calculations will be that portion of the work that must be done sequentially. If it takes longer to divide up the work than it does to do the work, the effect of distributed computing will be dramatically reduced. The Wikipedia article has the math, but the idea is very straightforward. Therefore, of course, while working on the upcoming Hadoop project, it'll be good to keep an eye on areas that can only be improved through faster sequential operation, rather than just dividing up the work.
The other morsel I took from this chapter is that I need to refresh my linear algebra...