Monday, February 25, 2008

The Future of Internet and the role of R&E networks

[Olivier Martin has put together a very good overview paper on the challenges facing the future of the Internet and the ongoing evolution and role of the R&E networks in that evolution-- BSA]


After a fairly extensive review of the state of the Commercial and Research & Education, aka Academic, Internet the problematic behind the, still hypothetic, IPv4 to IPv6 migration will be examined in detail. A short review of the ongoing efforts to re-design the Internet in a clean-slate approach will then be made. This will include the National Science Foundation (NSF) funded programs such as FIND (Future Internet Network Design) [1] and GENI (Global Environment for Network Innovations) [2], European Union (EU) Framework Program 7 (FP7), but also more specific architectural proposals such as the publish/subscribe (pub/sub) paradigm and Data Oriented Network Architecture (DONA) [3].


What is slightly surprising is that, despite the fact that the need for on-demand, i.e. switched, circuits has not been clearly established, somewhat overdue efforts are spent on developing various Bandwidth on Demand (BoD) middleware in Europe and North America, e.g. Autobahn , DRAGON , ESLEA , JIT , OSCARS , etc. Fortunately, the DICE (DANTE, Internet2, CANARIE, and ESnet) Control Plane working group is actively developing an Inter-Domain Controller (IDC) protocol, based on ESnet’s OSCARS technology. “As a result of both the DRAGON and DICE collaborations, Internet2 has recently released an early version of a turn-key dynamic networking solution, called the “DCN (Dynamic Control Network) Software Suite” which includes IDC software and a modified version of the DRAGON software. Deployed as a set of web services, IDC software ensures that networks with different equipment, network technology, and allocation models can work together seamlessly to set up optical circuits”.


7 Tentative Conclusions
The Internet has ossified. A clean-slate re-implementation is unlikely in the medium to long term (i.e. 7-10 years). However, some new ideas may find their way into the current Internet. The most urgent problem is to solve the explosion of the routing tables which is endangering the growth and the stability of the Internet, but this should be fairly easy to solve as the number of actors, i.e. suppliers of core Internet routers, is fairly small (i.e. Cisco, Juniper).
The next most urgent problem is the exhaustion of the IPv4 address space. Strangely enough, this is not seen as a high priority item by many major ISPs! however, IPv6 looks unavoidable some day, if one adopts the “conventional” view that all Internet capable devices, e.g. mobile phones, home appliances, RFIDs, etc., must be directly accessible, but, is this really desirable or even sound? NAT like solution, even so considered as “kludges”, are therefore very likely to flourish and even to slow down considerably, if not prevent, the deployment of IPv6. This process should culminate with the standardization by the IETF of NATs. Last but not least, one cannot exclude the possibility that IANA, will allow the RIRs to go to an IPv4 “Trading Model”, thus considerably extending the lifetime of IPv4 and also facilitating the migration to IPv6 by granting much needed additional time. An ongoing problem is the proliferation of security threats and the associated “degeneracy” of the Internet but the time horizons of the clean-slate Internet architects and the Internet Service Providers are so different that one must be prepared to continue living with it!
More programmable network devices, e.g. routers, multiplexers, should become available, however, “Active Networks” technology is unlikely.
Last Mile, affordable, broadband access, including Campus networks will remain very challenging and fast evolving technology.