Mushkin is good because they're respectable, and their speciality is memory. PC133 means the SDRAM runs at 133MHz, VC stands for Virtual-Channel, and its an optimization of normal SDRAM which makes it perform better, but its support and supply is very limited, not to mention its price is high, so don't worry about that, just get normal PC133 SDRAM.
Simply put, CAS measures the memory access latency, i.e. the time it takes(in memory clock cycles) to access memory, so the lower the better. However, performance benefits are usually negligible.
As you'v probably guessed, usually the more memory you have the better. Right now the standard is still 128MB I believe, but if you've got the budget, no harm getting 256MB and making use of it. There'll be a gain in gaming performance, but its really subjective whether its noticable.