Increasing router availability through virtualization

Abstract

Router high availability is an important challenge in network management. In fact, the breakdown of a router can be expensive for the network, as it can lead to congestion and to the loss of connections or packets - which, in turn, translates into loss of money. Vendors usually achieve carrier-grade availability by having standby routers ready to take over when a failure condition hits the active router(s). In such redundancy scheme, it is needed to duplicate practically every hardware component (e.g. the router switching processors and line cards), what makes the final high availability router very expensive. In this paper, we propose a redundancy model that tries to increase the overall router availability - by exploiting virtualization -, while also reducing the final device cost - by not having to duplicate every piece of hardware to obtain the desired availability level. The analytical results obtained show the proposed approach was able to achieve the intended goals.