AGOCG logo
Graphics Multimedia VR Visualization Contents
Training Reports Workshops Briefings Index
This report is also available as an Acrobat file.
Back Next Contents

12.4. Summary

12.4.1. Network Evolution in the Short Term

While TV networks have existed for sometime, data networks are only just beginning to approach their capabilities. Distribution of multimedia in the same way that TV is distributed is simply not possible on existing data networks. These networks have relied on the telephone system, and a few high capacity links between centres. The 'Information superhighway' is all about distribution of multimedia. The problem is one of speed and cost. If multimedia contains video, then a VHS type display can be made in real time if the transmission link can carry 1.5 Mbps of data. At present this can only be provided by :- So for the present most of us will have to use a low end alternative - ISDN and an Ethernet LAN. Ethernet can support less then 10 Mbps and ISDN can transmit data at 128 kbits/sec.The ISDN is adequate for live person to person video conferencing, and data transfer for many LAN applications.

There are two schools of thought on the use of multimedia over LANs. One is that ways can be found to transport multimedia over traditional LANs. The other looks to new LAN and WAN technologies, and argues that conventional LANs are unsuited to multimedia.

The reality is that some real time multimedia applications will continue to suffer performance restrictions when operating over LANs, but some applications will become available. Multimedia applications which are less demanding of bandwidth and which do not operate in isochronous mode will be more easily carried over traditional LANs.

To provide for high capacity on local and wide are networks there will be a period of considerable competition between switched Ethernet, isochronous (Iso-ethernet) Ethernet and local ATM solutions. Suppliers will attempt to provide equipment and adapters at prices which will encourage users to upgrade their LAN infrastructure.

For wide area networking narrow band ISDN will play an important role for some time. ISDN connections will replace the provision of analogue PSTN lines, enabling data calls to be placed worldwide. Initially there will be applications developed for ISDN use, which will then incorporate Iso-ethernet and local ATM interfaces. The ISDN network and ATM network will operate alongside each other, but once the ability to connect through from an ISDN point to an ATM end user appears the networks will begin to converge. ISDN applications and connections will serve as feeder applications for the higher speed ATM backbone.

LAN emulation over ATM will assist high speed connections between networks. As experience of ATM use develops and local ATM becomes more widespread applications designed for ATM use by users on with a PC or Mac type platform will become available.

For users in the academic community the twin criteria of cost and speed will continue to be determining factors.

12.4.2. Future Networks

The development of computer communications has been led by the very different environments of wide area networks and local area network. Wide-area networking is high cost, low speed, high error rate, large delay. The design of the switch was dictated by the need to manage and allocate the bandwidth effectively. Both conventional circuit switching and packet switching are difficult to implement at higher data rates. In most current designs for local area networks, where bandwidth is not expensive, simplified addressing and switching has been employed at the expense of the effective use of the bandwidth. In particular, networks such as Ethernet use broadcast as the normal distribution method, which essentially eliminates the need for a switching element.

Fibre optics has made high-speed information transmission possible. The problem is how distribute or switch this information when the intelligence needed to make switching decisions is operating in comparatively low speed electronics. For optical switching to be possible the switch must use very simple switching logic, require very little storage and operate on packets of a significant size.

For example, at a gigabit, a 576 byte packet takes roughly 5 microseconds to be received so a packet switch must act extremely fast to avoid being the dominant delay in packet times. Moreover, the storage time for the packet in a conventional store and forward implementation also becomes a significant component of the delay. Thus, for packet switching to remain attractive in this environment, it appears necessary to increase the size of packets (or switch on packet groups), or specify the route at source.

For circuit switching to be efficient at high speeds, it must provide very fast circuit set-up and tear-down to support the bursty nature of most computer communication. For long distance routes this is difficult because the propagation delay is greater than the transmission time at high data rates.

The choice of switching technology determines its performance, its charging policies, and even its effective capabilities. As an example of the latter, a circuit-switched network may not provide strong multicasting support. Since high speed networks will be built from point-to-point fibre links that do not naturally provide multicast/broadcast it is difficult to predict whether multicasting can be provided as part of a network protocol, or should be supported by higher layers only.

In the future, the host may see the network as a message-passing system, or as memory [Leiner88]. At the same time, the network may use classic packets, wavelength division, or space division switching. Future network protocols will need to provide a secure connection independent of the networks for applications to use.

Hitherto the concept of layering has allowed dramatic variation in network technologies without requiring the complete re - implementation of applications. Unfortunately, the layer interface designs are all organised around the idea of commands and responses plus an error indicator. For example, the TCP layer provides the user with commands to set up or close a TCP connection and commands to send and receive data. The user may well "know" whether they are using a file transfer service or a video call, but can't tell the TCP. The underlying network may "know" congestion is limiting the throughput to levels too small for acceptable video, but it also can't tell the TCP implementation.

The implementation of Quality of Service requests flowing in one direction from application to network are a partial solution. It would be more useful if a two way dialogue could be established between applications and the network(s) in a dynamic fashion. For instance if a World Wide Web application finds it is running very slowly over a congested trans-atlantic connection it could request a dial up ISDN connection as an alternative route or a higher class of connection from an ATM switch. Implicit in managing this request is the presence of a more controlled allocation of resources. When data calls are mixed with new services such as video streams unless these are separately recognised and controlled, there is little reason to believe that effective service can be delivered unless the network is very lightly loaded.

As networks get faster, bottlenecks move into the host. The speed of the asynchronous port on a PC is a classic example, when connected to ISDN at 64 or 128 kbps. Network adapters have a crucial role to play. The future network adapter may be viewed as a memory interconnect, tying the memory in one host to another. The integration of the network adapter into the operating system with the the transport level will be implemented largely, if not entirely, in the network adapter, would provide the host with reliable memory-to-memory transfer at memory speeds with a minimum of interrupt processing, bus overhead and packet processing.

Back Next Contents

Graphics     Multimedia      Virtual Environments      Visualisation      Contents