Survey: Live Migration and Disaster Recovery over Long-Distance Networks

摘要

We study the virtual machine live migration ( LM) and disaster recovery (DR) from a networking perspective, considering long-distance networks, for example, between data centers. These networks are usually constrained by limited available bandwidth, increased latency and congestion, or high cost of use when dedicated network resources are used, while their exact characteristics cannot be controlled. LM and DR present several challenges due to the large amounts of data that need to be transferred over long-distance networks, which increase with the number of migrated or protected resources. In this context, our work presents the way LM and DR are currently being performed and their operation in long-distance networking environments, discussing related issues and bottlenecks and surveying other works. We also present the way networks are evolving today and the new technologies and protocols (e.g., software-defined networking, or SDN, and flexible optical networks) that can be used to boost the efficiency of LM and DR over long distances. Traffic redirection in a long-distance environment is also an important part of the whole equation, since it directly affects the transparency of LM and DR. Related works and solutions both from academia and the industry are presented.