Egalitarian stable matching for VM migration in cloud computing

Virtual machine migration represents a new challenge to design efficient and practical migration algorithms that work well with hundreds or even thousands of VMs and servers. In our previous work, we advocated the use of a general stable matching framework from economics to solve networking problems, and showed that it offers unique advantages compared to utility-based optimization, using a simple procedure of deferred acceptance. In this paper, we apply this framework to the VM migration problem, and propose an egalitarian approach that finds a stable matching fair to both VMs and servers, instead of favoring either side as a result of the deferred acceptance procedure. Such an egalitarian stable matching that minimizes the total rank sum of the outcome is shown to be a sensible notion of fairness under mild conditions, and through simulations is demonstrated to offer superior results.