Data deduplication for replication-based migration of virtual machines

Abstract

A migration system provides data deduplication for migrating virtual machines across a network. The system maintains a centralized deduplication dictionary storing key-value pairs of identifiers and corresponding virtual machine data mapped to the identifiers, which are transferred as part of a migrating virtual machine. The migration system can store at least a portion of the virtual machine data, in intermediate storage implemented on one or more storage devices. The intermediate storage is queryable using a hash-generated key of the stored data. After an initial transfer of data from a source computing environment to a target computing environment, a requesting computing environment can subsequently initiate a data transfer by first querying the migration system to determine if the target data has already been transferred and stored in intermediate storage. Instead of resending the virtual machine data from its source computing environment, the migration system initiates the transfer from the intermediate storage.