Slashdot Mirror
Whatever Happened to Internet II?
Julio writes "Whatever happened to the Internet II?
This cpnet story says
'There is a computer science networking instructor at the University of Wisconsin, which is an I-2 institution, that is collaborative teaching a course at a college in Japan on computer networking. The students in Wisconsin were able to hear an expert on networking who just happened to be in Japan and they weren't constrained by being in Wisconsin,'" Apparently 150 colleges are hooked to I-2 already, and it's growing steadily -- and quietly.
My super-precise calculations indicate that 50% of the world's bandwidth is currently wasted on college students playing quake. The other 50% is college students downloading pr0n. Therefore, if all the colleges of the world start using internet2, we will all have super fast connections too since they will not be downloading pr0n and playing quake on our internet!! yay internet2!! ;)
Seriously though, this technology seems pretty impressive. The current internet was so poorly designed. Its barely even salvageable. We need a new system designed from the ground up to be fast and efficient. Too bad internet2 probably wont be that for home users for many years to come. Even if we all get DSL, our packets will not be routed in a reasonable manner. Traceroute your connection to your favorite websites and you'll see what I mean.. you never know when your packets will reach the next bottleneck.
My school (Johns Hopkins) is part of the internet 2, and unlike what a previous poster said, all computers have automatic access to it. I don't have access to my computer at school now (the school shut down the network for y2k...) but if I traceroute a host on any member network, for example www.mit.edu, traffic goes through through vbns.net routers (the i2 routers). (Normal traffic doesn't go through the vbns.net routers.) I haven't done much with it, but friends have reported rates of almost 1MB/sec to other member schools.
Some links:
vbns network map
Internet 2 connected schools
People might be interested in reading George Gilder's "The Coming of the FibreSphere". Basically he calims that you can substitute mass cheap bandwidth for switches (which being electronic only add latency) creating a design of dark fibre with all the intelligence at the peripheral. Now while this may appeal to customers, certain telcos suddenly find themselves in the commodity bandwidth business with nothing to support their big expensive time-based, distance-function bills. Guess what their natural response is? How can they justify the $n per megabyte when they can't control the marginal costs and thus segment the market by imposing deliberate latencies or constraints. Remember that in the IT industry, the value migrates to the complex and difficult areas (e.g. CPU, complex software) so with companies investing in voice-activated smart phones, they lose control unless they can corner any new markets and introduce delaying tactics. Why bother with switching when you can tune to 1 of thousands of fibre frequencies, especially when you can't use more than a few hundred home shopping categories anyway. Anyway, the hope is that by giving the smart universities some taste of what is possible, they will develop bandwidth-hungry applications that will drive consumer demand and thus make large-scale cost effective infrastructure investment. Life will be interesting.
:-).
It is rather interesting that the base human desires seem to dominate new technology. I've heard an urban ledgend that the vibrator was the third patented invention that used the new minature electric motors (after sewing machine and something else I can't recall at the moment), the porn industry is leading with DVD and the porn sites (and gambling) are one of the few profitable internet enterprises. Not sure whether this is a commentary on applied technology or human nature though
LL
Yup, that's it. Higher bandwidth, reserved for the people who developed the internet in the first place, in the pre-"public" days.
There are good reasons to have a segment of the internet (or whatever the fsck you want to call a bunch of machines connected by fiber optics with the purpose of sharing data) reserved for academia. One, it was the academics that developed it to begin with. Two, do you have any idea the amount of data people in academia need to transfer? Probably not... I do.
When I was working on my PhD, I was performing large, 3D simulations on one of the Cray's at the NCSC. The data files for those simulations totaled 4.5 GB per simulation. For completeness, I had to run several of these simulations. Can you imagine how long it would have taken to download all that data on the same lines that all the "commercial users" use? Weeks or months. Even with I-2 access, it took me a couple of hours to download data from each of these simulations.
Now, before you start whining about preferential treatment of academics, ask yourself this question: Does (pick your {least?} favorite average user) *need* to download GBs of pr0n and mp3s? Does the academic's job depend on being able to access GBs of scientific data? My answers are No! and Yes! in that order...
My $1.47
Eric
Well what is internet-2?
... Distance Learning, TeleMedicine, TeleMaintenance, Collaboratiive Research and Science, VTC, ... Knowledgebase - automated Data collection, manipulation, interpretation, distribution, ....
... I'm no expert (just interested).
"Reality is a self-induced hallucination."
To some in education it is an education internet.
To some in the military it is a DARPA internet project.
To some in business it is the future of the B-B eBusiness world internet.
I can not speak for internet-2, but I can say what I think ("AFT").
Internet-2 is a project with the intent to provide a developmental space for "Things to come.". Significantly greater bandwidth, vastly improved bandwidth utilization, resource management and control, and most importantly a truely enhanced functions and features rich environment for all internet users.
Internet-2 is advancing, discovering, and developing what will pe part of the future internet
Anyway this is a little of the way I look at the intent for the future internet. PLEASE, do not make the mistake of interpreting anything I said as a "1984 - Big Brother" concept. The future internet will be for the people just like today's internet. Dang Good Stuff on it's way to US folks and y'all. I just read about this stuff
WISE-YES and APES
Unaccountable leaders are masters, and unrepresented people are slaves. How do US and EU fare?
The Internet was rather well designed if you take account of the fact that it was initially intended only to share data and provide remote access between academic institutions, and not many of them at that. Now it has grown beyond all imagining, yet it still works using essentially the same protocols - a mark of good design IMO.
m - this has useful links at the end. For IPv6, see http://www.ipv6.org/.
As for I-2, it will be IPv6 based, but contrary to popular opinion IPv6 is not automagically faster or better than IPv4 - while v6 has many nice features such as autoconfiguration, auto-addressing, large address space, etc, there are very few features designed to make things go faster. All the technologies listed below apply equally to IPv6 and IPv4:
- MPLS - Multiprotocol Label Switching - allows administrator fine tuning of the routes taken across the network, e.g. to balance loads over the whole network, can also be used for VPNs and QoS.
- DiffServ - Differentiated Services - lets you assign a priority level to every packet (e.g. gold, silver, basic) and make gold packets get some guaranteed bandwidth or lower latency, hop by hop. Easy to deploy, does not give cast iron QoS guarantees.
- IntServ and RSVP - Integrated Services and Resource Reservation Protocol - lets applications request a certain QoS (bandwidth, latency, etc.) end to end across a network. Harder to deploy across a network, and has scalability problems, but these are gradually being addressed and it does give end to end guarantees.
There is one neat feature in IPv6 that supports RSVP - it's called the Flow Label, and is basically a number that is assigned to all packets in a given 'flow' (e.g. a video session). By assigning this number, RSVP routers after the first one in the path can go somewhat faster since they only need to look at one field rather than checking src/dest IP addresses/ports.
Windows 2000 includes many QoS features, particularly RSVP/IntServ and DiffServ, but not IPv6. RSVP is available for Linux, IPv6 is available in early form, and the Linux-DiffServ project is one of the most advanced implementations of DiffServ that is publicly available.
For more information on QoS, see http://www.qosforum.com/docs/glossary/glossary.ht
Of course, the ability to send traffic over big fat optical pipes is available to v4 and v6. However, the cost of ASICs probably dictates that gigabit/terabit routers may only support IPv4 for some time, until v6 becomes more widely deployed. However, I-2 may well be using early versions of v6 gigabit routers.