Immediate Recovery Of Virtual Machine Using Deduplication Device and Snapshots

摘要

Example apparatus and methods use two types of storage to improve boot time associated with recovering a virtual machine running on a hypervisor. The first storage is deduplication based and the second storage is not deduplication based. Example apparatus and methods may read data from the first storage to support operations immediately upon the VM booting and until a full recovery has been completed. Example apparatus and methods may also write a snapshot to the second storage and then support random input/output for the virtual machine from the snapshot in the second storage immediately upon the VM booting until a full recovery has been completed. Having dual devices facilitates using deduplication storage for deduplication-centric I/O while non-deduplication storage is used for random I/O while transfer of a recovery image from a backup device to a primary device associated with the VM is completed.

主权项

1. An apparatus, comprising:
a processor that runs a hypervisor; a memory that stores information associated with one or more data storage devices available to the hypervisor; a set of logics that control input/output with the one or more data storage devices for a recovery process associated with recovering a virtual machine (VM); and an interface that connects the processor, the memory, and the set of logics, the set of logics comprising:
a first logic that provides read access to a backup image from which the VM can be recovered, where the backup image is stored on a first data storage device;a second logic that provides read/write access to a snapshot associated with the VM, where the snapshot is stored on a second data storage device, where the second data storage device is a device other than the first data storage device, anda third logic that controls the hypervisor during the recovery process, where controlling the hypervisor during the recovery process includes:causing the hypervisor to create a new instance of the VM,causing reads associated with creating the new instance of the VM to be directed to the first data storage device,causing reads and writes generated by the new instance of the VM to be handled by the snapshot,causing the backup image to be moved from the first data storage to a primary data store associated with the new instance of the VM to create the recovery image without delaying reads associated with booting the new instance of the VM from the first data storage device and without delaying reads or writes to the snapshot, andupon determining that the recovery image is ready for use by the new VM, redirecting input/output for the new VM to the recovery image.