understanding vmstat

I'm trying to understand some of the performance monitoring tools in Solaris, below are a couple of vmstats from different systems.  If I understand everything, 'server B' is actually busier than 'server A' since 'server B' has more context switches, even though it has less interrupts.  I'm also not sure I believe the system idle time is 100% idle with a runnable thread on the queue.  I seem to remember reading something somewhere that said 'id' could go a little goofy in some situations and report either 99 or 100 but not truly be that idle.

Another question, how do you determine if your css or int are too high on a server and is there a correlation between int's and csw's or a way to drill down into why you have high csw's or int's?