Distributed hash group-by cooperative processing

摘要

A kind of system of parallel and collaborative process data is provided, wherein mediators handle and one or more agent processes and data processing function, which partially leaves loading. The agent process reads and handles data and added up partial results. Each broker program is handled in response to the content of the data of statistical collection, the process of the processing coordinator of returning part result. These steps are iteratively repeated, are completed until processing has. In specific applications, the performance of data processing system is improved, by accelerating database group-by inquiries. Host Central Processing Unit (CPU) (CPU) and input/(I/O) processor (IOPs) are distributed between the group-by operation processings. Substantially, IOPs group-by requests are sent to the selection inquiry for executing one group of disk block (range) and predicate tuple. The entry that the IOPs establishes Hash table has group-by elements and operation and aggregation capability (for example). The IOPs search domains extract record, and selection record is specified using the predicate, into the element in Hash table, if it has existed, and executes corresponding set function.