It seems I may owe a partial apology to the doomsayers who claimed the Internet “died” the other week. No, you still were blatantly wrong as evidenced by the web’s rapid rebound after its “demise,” but the idea that the Internet is broken may not be entirely wrong. At least that’s the claim being made by Lawrence G. Roberts, the man who led the team that developed the ARPANET, the forebearer to the modern Internet.
Actually, “broken” is not the most accurate term, as Roberts himself admits, but in a featured article on IEEE Spectrum, the flagship publication of the IEEE, he explains that the system he helped design 40 years ago is not optimal for new methods of sharing data, such as P2P and streaming media.
Roberts explains the problem in fairly straightforward language: routers are designed to handle data as packets but media flows so quickly and P2P streams through so many locations that routers work much harder to process them, resulting in their excessively disproportionate bandwidth consumption. His solution is to treat this data as flows, processing only the initial data packet then applying that processing to the rest of the data that follows behind it.
I liken it to the way energy travels as both particle and wave. Data traveling over a network is often viewed as parcels delivered by postal service. That was fine early on when data consisted of fixed files, but now it’s advanced beyond packet delivery (particle) and includes streams or flows (wave) that move fluidly and are not individually encapsulated. Considering data travels as energy through the Internet (electricity through copper lines and light through fiber optics), treating it like energy seems like the most logical approach.
What I found most encouraging is that Roberts has, through his company Anagran, developed technology to deal with data as flows and eliminate the excessive bandwidth consumption of streaming and P2P, as well as deal with the rising consumption of cloud computing. The potential impact on mobile connectivity is tremendous as it could help loosen bandwidth caps and restrictions on data usage. On the flip side, flow management also offers the ability to selectively shut down certain flows. Such is the nature of power.
Hat tip to Ars Technica.