Live migration of virtual machines is a primary method of load balancing on high performance servers in cloud computing. However, this involves a certain degree of penalty to the user. Minimizing this penalty is an ongoing research problem. We tested multiple VMs with different benchmarks for their time consumption in live migration. Based on the different types of workloads, we were able to detect patterns in live migration that enabled us to classify the different permutations in which machines should be migrated. This was mainly based on the CPU and memory usages of the respective VMs during benchmark execution.