Estimation, analytical modeling and synthetic load testing are the three best ways to predict workload capacity requirements. Up front planning can help you efficiently manage those workloads.
Of the three options mentioned above, you should have a good understanding of what will work best in your environment (virtual and physical, homogeneous and heterogeneous, simple and complex). Remember, managing your environment is part of your journey to simplifying your work.
Read an excerpt from the white paper below and check it out for yourself.
Tetris-like blocks symbolize the irregularities among different workloads. A simple workload with a moderate need for resources would be represented by a basic two-piece block. Higher needs for resources and higher complexity would cause the block to expand in various directions.
The larger and less symmetric the blocks get, the harder it is to combine them. An inability to combine the blocks translates into workloads that are starved for resources and can't be migrated to another host. All of a sudden, two of the key mechanisms that enable flexibility in virtualized environments become unusable. And even if you were able to combine them, large asymmetric blocks will most likely lead to white space fragmentation and lower resources utilization than you calculated and planned for.
Is this something you've dealt with? Let me know which option worked best in your environment and why?
Enjoy the journey.