MBONE: Multicasting Tomorrow's Internet
Over time, more and more sites will begin routing multicast packets in their native fashion, with true mrouters. As more sites gain experience with this task, the number of sites using IP tunneling will decrease. IP tunneling using mrouted is a clever hack -- but compared to true multicasting, it is a relatively inefficient system.
Another problem arises from this transition away from mrouted: by using mrouted, we've grown accustomed to its advanced features, such as the support for mtrace and pruning. Currently, most native multicast routers do not support these features. Although multicast routers will eventually incorporate these features if they are to take over for mrouted, this is not the case today; as a result, the deployment of native multicast routing has been slow.
The next biggest problem deals with the bandwidth of the Internet. The MBONE was designed to use 500 Kbps to transmit multicasts, which is an adequate maximum bandwidth for now. As the MBONE becomes more popular, however, the 500 Kbps maximum bandwidth will not be sufficient for handling the increased demands placed on it. Not only will the number of events that conflict with each other increase, but the needs of the people using MBONE will also grow. One day users will certainly want to transmit full-motion, full-screen video via the MBONE, and this capability will require lots of bandwidth -- more than the 500 Kbps limit imposed on the MBONE today.
This last problem is relatively easy to solve -- the simple solution is to give it time. Five years ago, the links that existed between the provinces of Canada consisted of 56-Kbps links. At that time, the United States had begun working with T1 links. Today, the links that exist between the Canadian provinces have been upgraded to 45-Mbps ATM (asynchronous transfer mode) links. The United States also uses ATM links, multiple T3 links, and now, interstate fiber-optic links. Also five years ago, an individual could connect to the Internet from his or her home at a speed of 9600 bps; today, that same individual can connect to the Internet with an ISDN link (128 Kbps) for about the same price. As time passes, the capacity of the links is increasing; conversely, the cost of these links is decreasing. At some point in time, bandwidth may cease to be a problem altogether.
Phone companies already have plans to bring video into the home, but the problem is trying to find an efficient way to get it there. One solution is to run fiber into the home, but currently, the cost of carrying out this particular solution would be astronomical. Another solution is to bring fiber-optic links to distribution centers (one per neighborhood or area) and share this link over the service area.
Whatever solution is adopted, real-time home video will one day be a reality. At that point, the MBONE technology will be so deeply integrated in our everyday lives that the MBONE will cease to exist as a separate entity from the Internet itself.
Another big problem with using the MBONE is of a more technical nature. Currently, the Internet's network protocols do not provide the necessary support needed for real-time service. This means that the MBONE cannot work as well as it should. Real-time traffic requires minimal delays between the transmitter and the receiver and low-packet loss rates. The Internet currently doesn't have the capabilities to ensure that real-time traffic will be delivered with minimal delays and low loss rates. Packet loss rates strongly influence the quality of service that you can get from the MBONE.
A partial solution to this problem is called IPng, or IPV6. IPng (Internet Protocol, Next Generation) is the next version of the IP protocol, and its principles have already been adopted. The transition to the new IP protocol is expected to take place over the next ten years. This new protocol will provide the MBONE with the provisions that it needs to support real-time traffic. To do this, the new IP protocol will be able to determine the real-time traffic's needs and then take that into consideration when routing the various kinds of traffic. Time-dependant traffic (such as a real-time multicast) can be routed quickly; less time-dependant traffic (like an e-mail message to a friend) can take the slow boat to its destination.
Another part of the solution is still to come, however. In the meantime, networking research is being done on the important issues that still plague our real-time traffic capabilities, and IETF (Internet Engineering Task Force) workgroups have been formed to develop standard protocols that would support real-time services on the Internet. In the future, we can expect these workgroups to solve problems related to such topics as admission control (refusing a request if the currently available resources are not sufficient) to packet classification and queuing (scheduling each packet for routing in the appropriate order of priority).
Another related problem is that the tools for managing a huge structure such as the MBONE are still being developed. Although some MBONE management tools are available, they do yet not form an entire MBONE management package. You can obtain these tools from a variety of sources, but you make have to go through a great deal of work to retrieve everything that you need to properly manage your MBONE site.
Again, this problem will be solved over time. The management tools will continue to evolve and become integrated with the MBONE infrastructure.