Capacity Planning

I am looking for some data that will help me do capacity planning.

How do i decide how many applications should be hosted on a particular server ?
WHat %age of CPU utilization is max before i plan to scale horizontally or vertically ?
If I am using clustering then when and based on what data do I plan to add more nodes to my clusters ? ect

I am looking for some unix commands and limits that I can observe before scaling out. Are there some formulas that will help me to arrive at these deciscions.

Please provide any data, document or url that will help me answer these question.

Thanks and appreciate your help .

pk