Highly reliable and scalable architecture for data centers

摘要

The present invention provides a highly reliable and scalable architecture for data centers. Work to be performed is divided into discrete work units. The work units are maintained in a pool of work units that may be processed by any number of different servers. A server may extract an eligible work unit and attempt to process it. If the processing of the work unit succeeds, the work unit is tagged as executed and becomes ineligible for other servers. If the server fails to execute the work unit for some reason, the work unit becomes eligible again and another server may extract and execute it. A server extracts and executes work units when they have available resources. This leads to the automatic load balancing of the data center.