A Novel Method for Calculating Query Hashes for Improved Query Grouping in Relational Database Management Systems

Database queries are stored and compared by relational database manage-ment systems as hashes, or short unique representations, of the original query text. This leads to cache misses and increased resource consumption by data-base engines when queries differing only in non-syntactic detail, or queries which are relationally equivalent, are presented to the query parser. We pro-pose a new method of structural query decomposition, transforming database queries into multidimensional adjacency cubes (MACs), allowing the codifi-cation of queries by structure rather than content as currently implemented. We build and test our solution, demonstrating superior query hash grouping to that currently offered by a leading relational database platform, and consider the applications of this new technique.