Until recently, many service providers maintained and operated separate legacy circuit-switched and packet-switched networks. Traditionally, voice services have been offered over circuit-switched networks, commonly known as Public Switched Telephone Networks (PSTN). On the other hand, connectivity between enterprises for virtual private network (VPN) data applications has been provided over packet-switched networks such as Frame Relay (FR) and Asynchronous Transfer Mode (ATM). Of late, many service providers are migrating legacy Layer 2 and Layer 3 services to converged Multiprotocol Label Switching (MPLS)-enabled IP networks.1 This migration toward a common multiservice IP/MPLS network is driven by the necessity to reduce the capital expenditure (capex) and operational expenses (opex) of both building and operating separate network infrastructures.
This chapter describes major sources of network failures and provides an overview of techniques that are commonly used to improve availability of IP/MPLS networks. In particular, this chapter outlines mechanisms for reducing network downtime due to control-plane failures.
Reliability and Availability of Converged Networks
For service providers, maintaining highly reliable and revenue-generating legacy service offerings is extremely important. So as much as possible, they are interested in migrating legacy services on to IP/MPLS infrastructures without cannibalizing revenue from these services. During migration, they also try to keep network downtime to a minimum (for example, in the order of a few minutes per year) to keep the cost of network outages in check. For example, a 1-minute network outage that affects 100 customers could cost a service provider several hundred thousand dollars.2 Therefore, it is not surprising to know that network reliability and availability rank among the top concerns of the most service providers. In short, high availability of IP/MPLS networks is a prerequisite to offer reliable and profitable carrier-class services. A well-designed network element, such as a router, facilitates the building of highly available networks and reduces the capex and opex associated with redundant network infrastructures.