Adaptive file allocation in distributed computer systems

Abstract

An algorithm to dynamically reallocate the database files in a computer network is presented. The proposed algorithm uses the best fit approach to allocate and delete beneficial file copies. A key problem of economical estimation of future access and update pattern is discussed and an algorithm based on the Gabor-Kolmogorov learning process is presented to estimate the access and the update patterns. A distributed candidate selection algorithm is presented to reduce the number of files and nodes in reallocation. The simulation results are presented to demonstrate both accuracy and efficiency of the proposed algorithms.