The Next Generation Internet: Scenarios and suggestions for Research Directions A White Paper submitted to the Workshop on Research Directions for the Next Generation Internet Joseph Fernandez Senior Consultant JNA Telecommunications Limited 16 Smith St Chatswood NSW 2067 Australia Telephone: + 61 2 9935 5591 Fax: +61 2 9417 7804 E-mail: joe.fernandez@jna.com.au Peter Stevenson Senior Consultant JNA Telecommunications Limited 16 Smith St Chatswood NSW 2067 Australia Telephone: + 61 2 9935 5537 Fax: +61 2 9417 7804 E-mail: pste@jna.com.au The Next Generation Internet: Scenarios and suggestions for Research Directions The Internet of the 21st century will continue to be basically what it is now and what it has been since its inception in the early 1970s - not one single network but a loose federation of interconnected networks. Some would say that the Internet's modus operandi borders on anarchy - agreement on standards through rough consensus rather than a formal vote, dissemination of other requisite technical information through folk lore, and the freedom for any and all organisations to connect to the Internet by implementing these technical requirements. In fact, this approach has served us well and the explosive growth of the Internet is a testament to the success to date of the process (or lack of process) that underlies Internet development. With anticipated continued strong growth based on this model, the danger is of the strengths being swamped by the weaknesses. We can envisage very different scenarios for how the Internet of the 21st century might develop. Scenario 1 Growth leads to severe bandwidth and reliability problems for Internet connectivity. The amount of unreliable, low quality content on the net increases much faster than the amount of reliable, high quality content. Information is pirated and the age of misinformation arises. Sites masquerading as authoritative ones attempt to make people believe all sorts of things in order to sell products, and often succeed. The landscape of the net becomes like something out of "Blade Runner" - lots of advertising, and little worthwhile information. Copyright and privacy legislation fail to keep up with the international nature of the Internet. Some organisations see a business opportunity here to provide a controlled reliable service. One or two global players, with the resources to deploy high speed backbones, institute a certification programmes for ISPs, and QOS agreements for end customers, and permit access only to a subset of sites, those that have been authenticated and certified. This leads to the formation of two Internets - the high grade and the low grade. Scenario 2 While bandwidth and reliability problems are encountered during the early growth phase, deployment of high speed Sonet/SDH and ATM backbones and research and development on protocol enhancements and switching and routing technology gradually lead to ubiquitous availability of fast, reliable Internet access. The Internet levels the playing field for commercial enterprises and leads to the development of a flourishing cottage industry. Small companies use the Internet as a cost effective tool to gain information, communicate with suppliers, advertise their products, and offer technical support to customers. Larger companies use it to coordinate geographically distributed development, manufacturing and customer service teams. Universities and schools provide distance education to remotely located students. Domestic activity on the net increases as new entrants into the carrier market take the Internet into more homes. Interest in the Internet continues to rise, and (unlike the case of CB radio) its popularity does not decline, owing to the diversity of things that can be done with and on the Internet. In the next phase, people use the Internet without being aware of it (for example, there are Internet-linked gaming machines in video arcades). Home appliances like stereos, TVs and VCRs are on home LANs connected to the Internet, allowing the latest software for them to be downloaded through the Internet. Suggestions for Research Directions In order to maximise the chances of the second scenario being realised rather than the first, research needs to be carried out in areas that can be broadly categorised as improving the Signal to Noise ratio on the Internet. Work needs to be carried out on content classification schemes, on intelligent agents for information retrieval, and on mail filters. Deployment of higher speed transmission links and of ATM switching in the backbone must be complemented with research on faster routing/switching techniques such as IP switching and label switching. The current nature of the Internet has been determined by "killer" applications. The two killer applications have been, first, e-mail (which gave us fast, easy, cheap communications regardless of location) and then the Web (which gives us ubiquitous access to information - but currently often slow and unreliable access). Logical extensions to world wide email and information retrieval over the Internet would be worldwide desktop conferencing over the Internet and distance education. (In some cases, timezones will constrain the use of these applications, but in many other cases effective use could be made of them.)This will need an Internet with a lot more transmission bandwidth, faster routing, and guaranteed QOS. The proposed NGI initiative with high bandwidth transmission facilities will, for the first time, provide the environment to run large scale experiments on using whiteboard type conferencing over the Internet to investigate the effectiveness of multicast protocols, QOS protocols, and fast routing/switching protocols in supporting this application. The goals will be to establish the bandwidth requirements, to determine where the bottlenecks are, and what modifications to protocols can give improvements. Currently we are going through a period of standardisation where most organisations are choosing TCP/IP as their network protocol. While the business community moves to this standardisation, the research and educational communities need to move away from it to the next generation of Internet protocols, a range of higher level protocols for specific applications that run over a core of IP or IPng. The Internet was originally created by initiatives funded by the US Department of Defence for computer-to-computer communications, not human-to-human communications via computer (e-mail) or computer-to-human communications for information retrieval (the Web). The original concept was that of linking computers together to work on common problems. That concept has not been implemented to any significant degree. However, experimentation with protocols has led to the deployment of the Internet in its current form. The most important guideline for Internet research is that this experimentation with a wide range of applications and high level protocols should continue, so that, as in the past, applications that are not currently envisaged can emerge.