The only correct way to do capacity planning is to plot application
throughput
with system utilization. Maybe you can run 40 Zones each with a webserver,
and a database Zone, or maybe 4. Or 1. It's completely workload dependent.
The system utilization data, without corresponding throughput and
That's cool, and I certainly respect your expertise in this area
(among many others when it comes to this stuff).
The problem we run into (more often than not) is that these
things are rarely linear. There's a knee in the curve lurking
behind a dark corner that can trip things up before we're at