Abstract
Resource allocation is an important issue in any shared computer system. Recently, dynamic multi-resource fair allocation became a hot topic in cloud resource management. The most recent studies assume that each agent submits infinite number of tasks. In this paper, we consider a more realistic dynamic environment, where every agent submits a bounded number of tasks. We design a generalized dynamic dominant resource fairness mechanism, and develop a combinatorial optimal algorithm to find a fair allocation. Experimental results show that the solution produced by the proposed mechanism is close to the optimal solution.