Whilst this table shows clear and distinct differences between Internet and Managed IP networks, there is a widespread misunderstanding that the Internet cannot be used to provide a delay intolerant service such as telephony because of its unpredictable quality of service. Service providers can add a layer of management to the Internet to improve quality of service and if quality is particularly bad may route traffic temporarily on circuit switched or managed IP networks. ITXC, a large wholesale international carrier uses the Internet in this manner for good quality telephony. Thus telephony can be provided either on managed IP networks or on the Internet with some additional management functions. This scenario is shown in figure 14.
Figure 14: Service provision and types of network  2.3.4 ATM
IP is not the only new network technology. ATM (Asynchronous Transfer Mode) is a link layer “small packet” that can either be used underneath IP or be used to provide virtual circuits to carry media without involving IP. Many of the new systems designed for telcos who want a next generation PSTN are ATM based and so the report includes these ATM developments (see sections 3.3 and 3.4).
3 Networking issues 3.1 Identification systems
The systems for identifying called and calling parties use names and addresses.
Names
A name is a “combination of characters and is used to identify end users. (Character may include numbers, letters and symbols)”5. An end user is “a logical concept which may refer to a person, a persona (eg. work, home etc.), a piece of equipment (eg. NTE, phone etc.), an interface, a service (eg. Freephone), an application (eg. Video on Demand), or a location”.
A name is distinct in function from an address, which “ identifies the specific termination points of a connection and is used for routeing”. Addresses are essential for communication as the end points always have to be identified in a way that can be used for routeing, but names are not essential. Names are added for some services to make it easier for users to identify the distant end-point or to provide an identification system that is independent of the structure of the networks or the current location of the called party.
There are two common naming schemes:
-
E.164 names (numerical strings) defined by ITU-T Recommendation E.164 – The International Public Telecommunication Numbering Plan. This scheme is a mixture of names and addresses. It started primarily as an addressing system but has migrated to become more of a naming system because location and operator portability are functions of names rather than addresses.
-
Internet names of the form “user@domain” defined by RFC 1035 - Domain Names - Implementation and Specification
The choice of identification scheme is related to the nature of the service because a service description needs to specify which type of name is used. This is important because:
-
users need to know how to identify their correspondents
-
the choice of identification system determines the set of potential correspondents that can be reached
-
interconnected networks need to have a common method of identifying communicating users
Addresses
IP addresses are divided, in principle, into two parts:
-
The identity of the network (the network part)
-
The identity of the interface attached to the network (the host address, which is the destination of the IP packet)
The range of addresses allocated to ISPs may be chosen to provide aggregation, ie ISPs that are connected to the same transit (backbone) operator may have adjacent allocations.
There are two versions of IP protocols, whose address formats differ significantly:
-
IPv4, a 32-bit address, which is used throughout the Internet but which is considered to be in increasingly short supply and whose allocations are being controlled carefully
-
IPv6, a 128-bit address, which is just starting to be used and should provide more than adequate capacity for the future.
IP address allocations are made in blocks to ISPs and are organised to be aggregatable so that traffic on a particular route is likely to have addresses in contiguous blocks. This is important to reduce the size of the routeing tables in routers where several contiguous blocks that the same route require only one entry. The size of these routeing tables is a potential bottleneck in the growth of the Internet as router technology is only just keeping ahead of the traffic growth.
ISPs normally allocate IP addresses to dial-up customers dynamically using the Dynamic Host Configuration Protocol6 (DHCP). Addresses are allocated from a pool only while the customer is logged-on. After logging-off the same address will be allocated to another user.
IP addresses are allocated to interfaces, but different communication streams using different protocols may share the same interface. These streams are differentiated using port numbers which are carried in the protocol (eg TCP, UDP or RTP) that runs on top of IP. The combination of an IP address and a port number uniquely identifies the source or destination of a stream of packets flowing between an application. Each application protocol has a “well known” fixed port number assigned to it plus a range of port numbers for dynamic assignment to communication streams.
3.2 Routeing 3.2.1 Routeing in circuit switched networks
In circuit switched networks, calls are routed step-by-step7. In most cases the signalling and the transmission circuit follow the same path and each switch decides which subsequent switch to route to8. The routeing decisions may be made using number analysis based on routeing tables stored within the switch with or without additional information from a database (intelligent network databases may be used for number portability or services where the number is not suitable for analysis with routeing tables such as freephone numbers). The point at which a switch determines the routeing information based on the exact location of the called party varies from case to case.
3.2.2 Routeing in the public Internet
In the public Internet, there is normally a two stage process for general routeing (the special arrangements for SIP and H.323 are described later). Most communications are established in a client to host mode, eg access to an email server or a web site, where the called host has a fixed IP address:
-
The first stage determines the IP address of the called host. The calling host (the client) uses the public domain name system (DNS) system to resolve the Internet name for the host at the distant end into a public IP address. This differs from step-by-step routeing because the calling end determines the distant end address whereas with step-by-step routeing this information may be determined at an intermediate point.
-
Packets are sent to the called host’s IP address and each router (packet switch) routes the packets according to routeing tables. Because the IP addresses use aggregation and reflect the connection topology of the Internet, the size of the routeing tables is kept to manageable proportions. Routers exchange information electronically about the ranges of addresses that they can reach, and this enables the routeing tables to be updated automatically or semi-automatically, saving administrative work.
This arrangement works satisfactorily where clients (eg the user’s PC) have IP addresses that are assigned dynamically by their ISP because the communications sessions are always established from the client to the host and the only incoming communications to the client come from a host that the client has first accessed.
The Internet is capable of supporting any-to-any communications such as telephony as well as client to host, but the problem of dynamic addressing must be overcome. There are several methods for determining a far end address:
-
The Session Initiation Protocol (SIP)9 (see later under “Protocols”)
-
The “Instant Messaging solution”
-
Additional communications
The “Instant Messaging solution”
This is a proprietary solution that has no established name but is widely used in instant messaging and portal services that provide voice communications to users of the same service.
A small piece of proprietary software is downloaded to the user’s PC and this software sends to the service provider a copy of the IP address allocated by the ISP when the user logs on, and another cancellation message when they log-off. The service provider maintains a database of the active IP addresses of its customers and uses this database to resolve called names to IP addresses.
Additional communications
Calling and called parties only need to determine each others’ IP addresses to establish voice communications. Where addresses are assigned dynamically, they can determine their addresses from their PCs and exchange these IP addresses by email or PSTN call and then establish voice communications over the Internet without any involvement of a telephony service provider.
3.2.3 Routeing in Managed IP networks
Managed IP networks at present are not interconnected with each other for telephony at the IP level, although they may have interconnection to the PSTN and to the Internet. The market objective of these services at present is to provide intra-enterprise services especially VPNs and interconnection between LANs at different sites. There may be some connections between customers of the same managed IP network. The absence of IP based interconnections is due to:
-
The inadequacy of the standards to support interoperability between different vendors
-
The early stage of development of the services and the focus on market segments that offer the highest margins
Managed networks normally use compatible products from a single vendor. These products are based on either SIP or H.323 with proprietary additions.
Share with your friends: |
