Hi Jim,
At the border, only "routes" (i.e., ASNs) are processed, not discreet user addresses. In the hierarchical Internet, the latter get sorted out and forwarded to the end user within the routing domain of the end user's individual ISP at the edge, as opposed to the core where border routers play.
re:
"I sorta lost the link on how OBGP would partially resolve the coming problem. Is it simply an effort to divert some of the traffic to an alternative routing system? Which routing system would grow, incrementally, until everyone is using it?"
Well, you're in good company, apparently. The OBGP draft never made it to full fledged RFC status because others share your view, for various reasons. For one thing, ostensibly the IETF won't consider it unless at least one major carrier will be willing to support looking into it. According to its proponents, however, it would actually be contrary to the interests of common carriers, so they can't get a vendor to back it, either, because without carrier support the vendor would be wasting their time in development on a product with a questionable future.
Btw, it isn't the intent of OBGP to solve the coming route lookup problem. On the contrary, from where I sit it would only exacerbate it. The intent, instead, is to allow free access to lambdas (by end users, however you want to define the term end users) that could be swapped, bartered, or otherwise brokered by various means that the author St. Arnaud is still sorting out, and he states, admittedly, that this is mostly a research project at this time and he openly acknowledges doesn't know where it might wind up.
Whether it does or not I haven't a clue, but I found the read in the January 2002 issue of the Cook Report on Internet provocative and a bit fascinating, nonetheless.
A summary of Cook's treatment can be viewed on his executive summary page at (I'll post a snip of it below, as well, although it doesn't do real justice to the actual article, complete with St. Arnaud's diagrams and the interview conducted by Cook, himself):
cookreport.com
----
January 2002 (10.10) NEXT Ca*net 4 Plans Customer Control of Lambdas
Regional Nets, Universities and Researchers to Be Able to Establish Wavelength Peering via Specially Tailored Switches - Making Telecom a Customer Owned Asset May Create Favorable Impact on What Appears To Be Surplus Fiber Infrastructure
<delete>
The Canadians have a vision that has been lost in the free market purity of the United States' political environment. Consider the theme of the CANARIE's 7th Annual Advanced Networks Workshop being held this week in Toronto.
"Following the recent release of the National Broadband Task Force report, there is increased awareness that a national broadband infrastructure serving all Canadian communities will be critical to Canada's ability to innovate. Information technology infrastructure will be one of the most important vehicles for promoting innovation and improving Canada's productivity, leading to increased wealth and economic growth.
Community broadband networks, provincial networking initiatives and national research backbone networks, are all part of the same continuum of providing a national innovation infrastructure.
In the future, research, education and innovation will not be solely a product of universities and research centers. A national innovation infrastructure will allow all Canadians in our schools, communities and businesses, no matter how remote or how distant, to be full participants in developing and using innovative applications and services.
New concepts involving "grids" and "eScience" are coming to assume greater importance in many branches of science. Some of this work could allow students in our schools, and eventually members of the public, to participate in basic research that otherwise they could only read about, thereby engaging them directly in Canada's "innovation culture." See also : canarie.ca
CA*net 4 is to be built in part on the premise that with huge amounts of fiber laid and very large amounts lit but still not fully utilized direct control of actual bandwidth can, for the first time, be placed into the hands of customers and then end users. According to Bill St. Arnaud: "Today networking is like computing was 40 years ago when the market was dominated by large mainframe computers. But in the 1970s the mini-computer came along followed by the PC which fundamentally changed our thinking of how to do computing. Computing became personal. The user was empowered to develop new applications and services that were not possible on a mainframe computer. With CA*net 4 we hope to move networking in the same direction as computing has gone in the last 30 years." Arnaud wants to turn the network itself into a customer owned and controlled asset.
"From day one we will be assigning ownership and control of individual wavelengths or STS channels to the GigaPOPs, universities and perhaps even individual researchers. They will be free to trade and swap amongst themselves and do what ever they want with those wavelengths. From day one we will also encourage these organizations to directly peer with each other and other international research networks using these wavelengths. But, initially the BGP optical peering will be done manually. Once OBGP is successfully implemented, it will allow theses organizations to automatically change the routing of the wavelengths and peering relationships without first contacting CANARIE. So rather than operating a traditional hierarchical IP network as many other research networks do today, CANARIE will only offer an aggregate IP network as an optional service for those organizations that don¹t need their own wavelengths."
When we asked his thoughts about the ways in which CA*net 4 might help the Internet build a viable business model for life after the current down turn, he replied: "I think the business and architecture model of the future will be of control and management moving increasingly closer to the edge, not only of the in terms of applications, but also in terms of control over the infrastructure. I think that one of the drivers for this will be as a consequence of the issues that of Larry Lessig has raised where content and distribution companies are trying to exert control over the Internet infrastructure to protect their intellectual property interests. Decentralization and minimizing control at the center will help thwart these challenges. We are already working on concepts with our industry and research partners to extend this concept of customer control of wavelengths all the way to the individual home. . . . .
In the future we see a physical network infrastructure that closely parallels Morpheus and other peer to peer networking paradigms. The end user will have a choice of whether they wish to subscribe any number of "walled garden" service providers. Or they may chose to physically connect to community networks to share files and data with high speed Gigabit wavelengths bypassing all traditional hierarchical service providers.
A lot of these concepts are still very speculative and unproven at this point in time. But what it does point out - is the critical role that research and education networks still play in the ongoing development of the Internet. . . . . Most importantly research and education networks will continue to play an absolutely critical role in exploring new concepts in networking which initially may appear very radical to the traditional telecom world.
-------
FAC |
