An efficient virtual machine checkpointing mechanism for hypervisor-based HPC systems

Abstract- In this work, we present the design of the Checkpointing-Enabled Virtual Machine (CEVM) architecture. A unique feature of the CEVM is the provision of an efficient mechanism for checkpointing Virtual Machines in hypervisor-based High Performance Computing (HPC) systems. Our goals are (1) to enable implicit system-level fault tolerance without modifying existing operating systems, applications or hardware, and (2) to minimize the space and time overhead needed to execute software that cannot tolerate faults. We accomplish these goals by leveraging hypervisor technologies that yielded tremendous reliability and productivity improvements in the HPC community. In this paper, we discuss the two novel protocols used in the VM checkpointing mechanism that emphasize on achieving efficiency as well as ensuring correctness. I