MECHANISM FOR CO-LOCATED DATA PLACEMENT IN A PARALLEL ELASTIC DATABASE MANAGEMENT SYSTEM

摘要

A database management system implemented in a cloud computing environment. Operational nodes are assigned as controller-nodes, compute-nodes or storage-nodes. The number of operational nodes, and their assignment as compute-nodes or storage-nodes can vary. Queries specify tables, with each such table assigned to a respective group of storage nodes. The number of operational nodes executing a given query may change by (a) changing the compute-nodegroup associated with a connection, or (b) adding or removing nodes from the compute-nodegroup; and/or distributing data from the tables among the nodes in a storage nodegroup. State information is maintained for each client connection, such that steps are executed assuming that the state exists. State changes are recorded during step execution for a given connection, so that subsequent steps from the same connection are executed with the state as left by a prior step, and so that subsequent queries on the same connection are executed with the state as left by the prior query.