CPU usage forecasting
Is there a method of CPU forecasting that is not a 100% correct, but at
least have a good probability of being correct?
For example, when I have x visits my CPU usage is n; when I have x+10000
visits, my n changes but this change is not linear. Therefore I can't just
take n/x*(n+10000) since this would just plot a linear change that won't
be near the real result.
Any ideas how to approach this task?
No comments:
Post a Comment