Geography aware virtual machine migrations for distributed cloud data centers

Cloud computing provides access to computing resources for a fee. Client applications and services can be hosted in clouds. Cloud computing typically uses a network of data centers that are geographically dispersed. The distance between clients and applications is impacted by geographical distance. The geographical distribution of client requests can be random and difficult to predict. This suggests a need to reconsider the placement of services at run-time through migration. This paper describes a framework based on software-defined networking (SDN) principles. It demonstrates algorithms that are periodically executed and determine candidate services to migrate as well as target data centers to migrate to and an evaluation. The evaluation shows that effectiveness of the algorithms.