High Availability Layers and Failure Recovery Timers for Virtualized Systems and Services

摘要

Highly available virtualized systems and services are required for applications that are sensitive to down time. The availability of virtualized systems and services needs to be on par with that for non-virtualized systems and services. However, high availability (HA) designs for virtualized systems and services are much more complicated than those for their non-virtualized counterparts due to the existence of independent multiple layers where each layer may have its own failure recovery mechanism. This paper defines a novel approach to the design of highly available virtualized systems and services. The recovery from failures is self-coordinated. There are no race conditions among layers.