A Preliminary Architectural Design for the Functional Hierarchy of the INFOPLEX Database Computer

摘要

Conventional computer architecture has shown its limitations in supporting large-scale information processing. Database machines which specialize in database functions have been suggested to alleviate the problem of increasing loads on these computers. The INFOPLEX database machine proposed by Madnick employs a highly-parallel computer architecture to achieve high performance and capacity. It contains two subsystems: the storage hierarchy and the functional hierarchy. This paper is about a design of the functional hierarchy, the subsystem of the INFOPLEX which performs database functions. As originally proposed by Madnick, the functional hierarchy is made up of hierarchical levels; each level is designed to perform certain database functions and is to be implemented by multiple microprocessors. The guidelines for identifying these levels are: (1) functional abstraction, and (2) pipelining of transactions. This paper attempts to turn this original idea into a detailed design by accomplishing the following: (1) Identification of functional requirements (i.e., features) of a generalized DBMS; these features are to be supported by the functional hierarchy; (2) Development of an integrated data model; (3) Detailed specifications of architectural levels, including their functions and implementation strategies; and (4) Pointing out future research directions. (Author)