This chapter describes what is meant by a carrier-scale IP network. It details the challenges in designing, building and operating a carrier-scale IP network, and lists the abilities required in a network to make it carrier-scale, e.g. scalable, manageable, performant and secure. Also described are the challenges for the future when developing an esoteric protocol or service, such as multicast or VoIP, into a carrier scale protocol or service. None of the challenges highlighted here should be underestimated in their capability to confound, and they are best met by applying the solid design and operations principles.

This chapter presents the technical details of how to design large IP networks. An ad hoc collection of routers and nodes grown organically could lead to an unstable and congested network. The principles presented here are therefore not optional for large network designs.The design of large Internet protocol (IP) networks, such as a Global Internet backbone, involve many design parameters, e.g. the number of anticipated customers, connections to other service provider networks, anticipated traffic flows, and routing implementation. In all these areas scaling is vital to ensure the network not only works and is stable initially, but can grow to accommodate more customers, increased traffic and more points of connection. The chapter uses the designs of BT's and Concert's IP backbones, which form part of the global Internet, to demonstrate the techniques employed. Particular attention is given to the approaches used to scale the networks while maintaining stability of the routing protocols employed.

This chapter summarises the key aspects of building a point of presence (PoP). It presents what must be considered when selecting and placing equipment in a site and describes the options available. This chapter could be thought of as describing the layer 0 of a network, the basic foundation upon which the rest of the network and all IP services are built.

There are a number of broadband access technologies which are capable of effectively providing access to broadband IP services. These technologies and their associated network platforms will continue to evolve. Broadband access is no longer just a 'fat pipe' - many systems have increased functionality, configurable features and intelligence. Hence issues such as service provisioning, interoperability and CPE auto-configuration are increasingly important challenges that need to be overcome for mass-market viability. The various broadband access technologies have their relative merits and all will continue to prosper as the insatiable customer demand for bandwidth continues. Many operators and service providers will use a combination of these technologies in order to best approach their target markets. In the UK some cable operators already use DSL on their 'siamese' cables which include coaxial cable (used by cable modems) and twisted copper pairs. ISPs are using both DSL and broadband radio in the USA to increase customer coverage. Radio systems in conjunction with in-building distribution via DSL are being used to target office blocks. Satellite is increasingly being examined to expand broadband coverage to rural areas. BT's current ADSL platform is well positioned to take advantage of the proliferation of broadband IP services. The platform has the capability to increase its functionality and evolve towards incorporation of APON and VDSL wireline technology.

A significant amount of the Internet's success can be contributed to the industrialisation, the creation of a carrier-scale solution, of dial access and its associated modems. It is the mass production of modems, the significant reduction of their cost, and the development of network systems which can handle hundreds of thousands of customers, that has brought the cost of Internet access within the reach of the majority of the population. This chapter introduces the major elements of a carrier-scale dial access platform and describes how they should be organised to deliver service.

This chapter describes the role of operations and the typical environment found at an IP operations centre. One of the key factors in satisfying customers is the quality of the service provided by the operations team. Two key ingredients support good operations - people and tools. BT has a long history of operating IP networks - its Internet operation was established in 1994 when its public Internet service was launched; however, it has been operating private IP networks for over 20 years. Over this time BT has gained valuable experience of operating IP networks in a rapidly evolving environment and has optimised its operations for best performance. The chapter captures the experiences gained and systems developed in the field of Internet operations. The most important factors in operating IP networks are: the organisation of operations; a clear understanding of the roles and tasks; a structuring of customer support into tiers or lines so that best use is made of people with valuable IP skills and customers receive best service; the development and deployment of suitable tools; a clear definition of the service surround required to maintain an operations centre.

Designing and building a large IP network is only a part of creating a carrier-scale IP network. The other significant part, without which no network could be called carrier-scale, is the operational support systems (OSS) - the collection of systems that are used to deliver and manage the network and its services. The effectiveness and the integration of the OSS has a direct impact on the quality of the services and the profitability of the network. In a carrier-scale environment, the OSS has to be a scalable system. With IP networks, new services are continuously being developed with the consequence that the OSS needs continuous redevelopment. This chapter describes the OSS, its relationship to the network and services, and methods for streamlining its design and future development.

This chapter explains a technique called 'multiprotocol label switching (MPLS) traffic engineering (TE)' which operates alongside the IP layer to provide this functionality and looks at potential implementation models. TE is simply about utilising the circuits available in the network to best effect. MPLS TE has been included in this section of the book as MPLS is one of the more recent IP protocols to have an impact on the design and operation of carrier-scale IP networks. It should be noted that traffic engineering is not essential for the operation of all large IP networks - a careful cost benefit analysis should be conducted before deciding to operate MPLS TE on a particular network. Some networks may find MPLS TE too complex and expensive to implement compared to the costs of wasted bandwidth.