Towards a Taxonomy of Cloud Recovery Strategies

Abstract

Recovering from failures of sporadic operations such as rolling upgrade or migration is complicated by the fact that the application being upgraded or migrated must continue to provide service. This means that recovery strategies for sporadic operations must include facilities for recovering from normal operations as well. As a step in deriving methods for recovering from failures in sporadic operations, we classify existing methods into four categories according to their purposes and the life cycle phase for which they are applicable. Not only does this taxonomy facilitate the research on recoverability of cloud sporadic operations but also it can help better understand the existing cloud recovery strategies.