Periodical auctioning for QoS aware virtual network embedding
Proceedings of the 2012 IEEE 20th International Workshop on Quality of Service
| Hi-index | 0.00 |
The goal of packet-switched networks has conventionally been delivering data to users. This concept is changing rapidly as current technologies make it possible to build network processing engines that apply intermediary services to data traffic. This trend introduces an extensive range of ways to develop and operate applications by allowing processing services customized for applications' needs at intermediate network nodes. The provision of such services is potentially a significant benefit for network users, as it can relieve individuals from the need to acquire, install, and maintain software in end systems to perform required functions. As such network services become more widely used, it will become increasingly important for service providers to have effective methods to configure applications sessions so that they use resources efficiently. On the other hand, it is equally important to design such extensible networks properly in order to ensure desirable performance of applications. This dissertation addresses these two key problems that arise in operating and provisioning extensible networks: configuring application sessions and designing extensible networks. First, we present a general method, called layered network, for the problem of configuring application sessions that require intermediate processing. The layered network method finds optimal configurations by transforming the session configuration problem into a conventional shortest path problem. We show, through a series of examples, that the method can be applied to a wide variety of situations. We also discuss how to configure applications that require reserved capacity and propose effective heuristic algorithms that are based on the layered network method. Second, for designing extensible networks, we generalize the constraint-based network design methods originally developed for conventional networks. We show how to incorporate arbitrary application requirements that are allowed by extensible networks in a flexible and general way. We also show how to extend the original framework to dimension both processing resources and link bandwidth. These results have been incorporated into a software package, the Extensible Network Planner (XNP).