Multiprocessor database machine architectures based on data flow analysis
Takayuki Baba, S. Bing Yao, Alan R. Hevner, Zhicai Shi · OSTI OAI (U.S. Department of Energy Office of Scientific and Technical Information) · 1983
A method for designing high performance multiprocessor database machine architectures based on data flow analysis is developed. The method consists of decomposing database functions onto replicated processors. This paper describes the decomposition process. Query evaluation strategies are decomposed into groups and mapped to an asynchronously pipelined multiprocessor system. An example illustrates this process in detail. A cost model is described to evaluate the effect of pipelining on the execution times of the strategy in terms of disc accesses, computations and data transmissions. Application of the cost model to various decompositions of the example query provides insights for the design of a multiprocessor database system. 14 references.