A decision support methodology for the file allocation problem in distributed computing systems

Abstract

A methodology using the decision-support-systems approach for solving realistically sized file-allocation problems in distributed computing systems is developed. The methodology allows the designer: (1) to deal with the operating-cost and response-time considerations simultaneously; (2) to adapt to changes in the strategic emphasis between the two objectives; and (3) to incorporate time-varying file-access patterns and cost structures. A solution procedure which integrates combinatorial optimization and simulation techniques is developed and validated by numerical experiments.<>