In the realm of distributed computer networks, resource allocation can be a game changer. This paper tackles the complexities of file allocation and capacity assignment in networks with a fixed topology. It emphasizes the critical balance between these elements through an average message delay constraint. The research seeks the most cost-effective ways to distribute information files across network nodes and assign capacities to network links, all while adhering to network delay and file availability requirements. A model is presented to solve this intricate problem, revealing its classification as a nonlinear integer programming challenge. Recognizing the computational limitations of deterministic techniques, the study introduces a new heuristic algorithm. This algorithm employs a decomposition technique, significantly reducing the computational burden. The results, derived from various network configurations, show that the algorithm, despite lacking theoretical convergence, yields practical, low-cost solutions. Ultimately, this work demonstrates the algorithm's ability to solve realistic network problems that deterministic techniques would find computationally infeasible. This has important implications for managing resources in large-scale distributed systems.
“ACM Transactions on Database Systems” publishes research on all aspects of database systems. This paper contributes directly to the journal's focus, by presenting an efficient heuristic algorithm for optimal file allocation and capacity assignment in distributed networks. This addresses a key challenge in database system design and management, offering a practical solution for resource optimization.