Flow and greedy algorithms of resource co-allocation in distributed systems

Abstract

In this paper, we propose combinatorial methods and algorithms for coordinating resources in distributed data processing systems with the structure of the executed job. We reduce the problem of resource co-allocation to finding formalized descriptions of the requirements based on flow algorithms and Gale optimal job decomposition into task clusters with greedy algorithms. We examine the conditions under which resources compositions with required properties exist and estimate asymptotic time complexity of the corresponding algorithms.