Xen Project User Summit

Virtual machine (VM) replication (replicating the state of a primary VM running on a primary node to a secondary VM running on a secondary node) provides software solution of application-agnostic hardware fault tolerance - non-stop service. Unfortunately, existing VM replication approaches suffer from excessive overhead, limiting their applicability and adoption. However, in practice there is no need for the secondary VM to perfectly match its machine state with the primary VM at all times, and this may be exploited to optimize performance. In this talk, we propose COLO (COarse-grain LOck-stepping virtual machine), a generic and highly efficient non-stop service solution, utilizing on-demand VM replication, as one possible solution.