Multiple hash tables based on access frequency

摘要

An access frequency is determined for each data unit in a data sharing environment employing a hashing scheme, either from an estimate or from a history of data accesses. The data units are then categorized into access frequency classes. Available hash entries are then allocated among the various access frequency classes, with more hash entries being allocated to relatively higher access frequency classes. Once the hash entries are allocated, the data units are allocated to the hash entries in a given access frequency class. In the simplest form, the data units are evenly allocated among the hash entries of a given access frequency class.