Multi-class resource sharing with batch arrivals and complete blocking

Abstract

A cloud provider hosts virtual machines of different types, with different resource requirements. There are bounds on the total amounts of each kind of resource that are available. Requests arrive in batches of different sizes, and are accepted if all the VMs in the batch can be accommodated; otherwise the request is blocked, with an associated loss of revenue. The trade-offs between costs and benefits are evaluated by means of an appropriate model, for which a novel solution is proposed. The applicability of that solution is extended, by means of a simplification, to very large-scale systems. Numerical examples and comparisons with simulations are presented.