Carrier Scale IP Networks: designing and operating Internet networks
Comprehensively reviewing the design, provision and operations of carrier-scale Internet networks, this book presents a good balance between leading edge technology and many of the practical issues surrounding carrierscale IP networks.
Inspec keywords: IP networks; 3G mobile communication; telecommunication network planning; telecommunication security; Internet telephony
Other keywords: voice over IP; personalisation; Internet security; location; wireless communications; telecommunications performance engineering; telecommunications network planning; 3G mobile networks; carrier-scale IP networks; wireless security; telecommunications network modelling
Subjects: Mobile radio systems; Telephony; Other computer networks; Computer communications; Communication network design, planning and routing; Data security
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 is an easy-to-read introduction to the technology and history of IP networks. It provides a condensed summary of IP terms and concepts essential to the understanding of the rest of this book. Those with a basic understanding of IP may be tempted to skip this chapter but some may find that it covers gaps in their knowledge. The chapter introduces the reader to IP networking, covering the fundamental principles of packet switching, through to the devices, protocols, technologies and applications used to create an IP network.
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 looks at the essential ingredient of the Internet - how the many different Internet service providers connect to each other so that anyone on the Internet can reach anything on the Internet. The general term used to describe how ISPs connect to each other is 'peering'. The implementation of an ISP's peerings will be a major deteminant of the performance of the services the ISP offers to its customers. This chapter is an introduction to both the key principles and the art of peering.
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.
Those building IP networks see two types of connection as key to them - the access connections, the cables, fibres or wireless systems that connect their customers to the PoP, and the core connections, the fibres that connect the PoPs together. This chapter describes the latter aspect, looking at today's and tomorrow's transmission technology. Utilising cost effectively the capacity of the fibres buried in the ground and being able to reconfigure the transmission network in a timely manner to meet rapidly growing customer demand are of interest chiefly to ISPs and carriers that have their own transmission network. However, the cost of installing and managing the transmission network is a fundamental cost that is inherited by the IP networks.
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.
This chapter describes wireless technologies that can be used for carrying IP data. UMTS is one of these technologies; it has had a high profile due to the cost of its licences but is also another technology that may change the shape of the Internet with its ability to bring IP applications to potentially billions of mobile terminals. Other wireless technologies with a potential to change our lives are wireless LAN and Bluetooth - wireless LAN may free many users from plumbing their offices and homes with cables, while Bluetooth will enable a new type of data network, a data network for our personal space.
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.
Satellites are raising their profile in the access network, migrating from their traditional role as trunk connections between core networks. This migration is being driven in part by fibre installation around the globe and in part by increasing demand for mobile and IP-based services to which satellites are well suited. This chapter provides an overview of those satellite technologies that can be used to deliver IP-based broadband services to customers.
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.
The challenge for an IP network operator is to have flexible processes which allow innovation and development which then allows lightweight processes for new advanced products, but within the same environment to have rigorous and efficient processes for well-established products. This chapter gives an overview of operational models and processes, using BT and Concert as examples. It should help the reader understand how to design and develop operations to be flexible and efficient.
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.
Networks need addresses to identify objects and locations in the network. The structure of the address is the single largest influence on the nature of the network. To fully understand a network you must understand its address structure. This chapter describes in technical depth the structure of Internet protocol (IP) addresses and how these addresses are managed.
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.
Carrier-scale IP networks can offer more than just Internet access, they can be used to deliver virtual private networks (VPNs), a service traditionally offered by frame relay and ATM networks. A VPN uses a shared infrastructure to carry traffic for multiple domains (e.g. different customers or communities of interest). Privacy (i.e. traffic separation) is provided using various techniques that can reside at layer 2 or layer 3 of the OSI model. IP VPNs, in particular, apply segregation at layer 3. The key factor in VPNs is that all traffic from one domain (or customer) shares the same infrastructure as other domains, leading to economies of scale. This is achieved while maintaining security and separation from other domains (or customers). The main driver to the development of IP VPNs is therefore cost. This chapter concentrates on the latest IP VPN technologies - tag switching and IP security (IPSec) - compares them to legacy systems, and positions them against each other.
