Flexible advance reservation models for virtual network scheduling

Abstract

Advance reservation services allows users to pre-reserve network resources at future instants in time. These offerings are already being used by a wide range of applications in scientific/grid computing, datacenter backup, and event broadcasting. Now most advance reservation algorithms are designed to schedule point-to-point connection requests. However, as new cloud-based services gain traction, there is a further need to schedule broader virtual network (infrastructure services) demands. These latter types are much more complex and comprise of an arbitrary number of virtual nodes mapped onto physical nodes and interconnected via a set of virtual link connections. This paper addresses this critical area and develops/analyzes several virtual network scheduling schemes, including partial provisioning strategies to improve network revenues.