Non-Periodic Check-Pointing for Fine Granular Retry of Work in a Distributed Computing Environment

摘要

Distributing work in a distributed computing environment that includes multiple nodes. An individual node can receive a work assignment, which can then be divided into a plurality of work units. A first work unit can then be distributed to a first worker node. At least a portion of the first work unit can be re-distributed to a second worker node in response to determining that the first worker node has experienced a failure condition with respect to the first work unit.