The document outlines a methodology for sizing virtual machines when migrating enterprise applications from non-virtualized to virtualized servers. It involves monitoring application CPU utilization and transaction times in the non-virtualized environment. Queuing models are then used to evaluate deployment scenarios and identify the minimum number of virtual CPUs needed for each application's VM to maintain acceptable performance levels once virtualized. The methodology aims to determine optimal virtual CPU allocations based on actual physical resource needs rather than guesses, to avoid overcommitting resources and performance issues.