Slashdot Mirror

While some level of regulation could help, the general tenets of the "net neutrality" movement lack a viable technical solution. They'll never happen. Now if you want to advocate for sensible regulations that will moderately improve things and call it "net neutrality" then that's fine by me, but "net neutrality" as the public understands it will never happen, or if it did happen the way some advocates define it, would have an overall negative impact on customers.

The only technical solution I could see that would allow a fully fair "neutral" network (and scale worldwide) would be an end-to-end ATM network. That would allow the customer's PC to treat each connection similar to a phone call and lock in QoS, with contention being resolved fairly by limiting the duration/number of these calls, and round-robin CAC with fallback to ABR on the client side. While this would be easier these days with most CPE/endstations well capable of being modified to support FATE/CIF it would be a project much bigger than IPv6, which we STILL haven't gotten very far on, plus it would require a complete retool of the entire network core, and eventual weening off of IP entirely to get rid of resource-hungry SAR problems. Never going to happen.

RSVP could be considered, but IP-level QoS mechanisms or even MPLS tag switching just don't have the oomph needed compared to finely honed/streamlined cell/circuit-based equipment. You'd have to totally rebuild the core for that, too.

As far as sensible regulation, really what needs to happen is oddly enough very similar to one of the health care reforms: require companies to spend a certain portion of their revenues from selling services on network improvements that improve the general state of the network. I'd add to that that any colo or special peering arrangements should be required to contribute to general expansion of the network in order to offset the hogging impact that latency inequality creates in the TCP stack.

>the internet was originally designed to run as a communications network in the event of a nuclear attack.

That idea was floating around but it wasn't what drove the MIT/DARPA work that turned into today's Internet:

http://www.ziplink.net/~lroberts/InternetChronolog y.html

http://www.ziplink.net/~wood/funny/snigglets.html

Snigglets (words that should be in the dictionary but aren't) were invented by comedian Rich Hall. I believe he came up with the idea on the ABC comedy show Fridays (a sort of copycat SNL), then wrote a couple of books on the subject.

I like the term he came up with for people who use the express lane at the supermarket, while carrying one or more items above the limit - expressholes.

Even at a modest 64kbps stream this would consume 32Gbps of bandwidth - that's THREE OC192's or, although the figures vary quite widely (Here's one), approximately the entire capacity of the "Internet" as it currently stands.

There are technologies that can handle this using a mere 64kbps in total (e.g. multicast) but they're not widely adopted/available (side note - why??)

You'd think an agency that can put someone on the Moon and vehicles on Mars would have the tech savvy to know off the top of their heads that they're dreaming!

Wierdly, some of them are still trading. Ziplink (ZIPL) is quoted at $0.0001 on NASDAQ. Their web site is still up, although most of the pages are bad links. Their last news item is "ZipLink, Inc. (NASDAQ: ZIPL), a wholesale Internet connectivity provider, today announced that the company plans to suspend its operations effective today, November 17, 2000."

Despite this, the stock is still tradable, and a few people trade it each day.

Hello?

This happens in ALL skilled professions where quality work output is dependant on ability and knowledge.

Think famous scientists

A reply on the same thread: For geeks, you slashdot kids display an embarrassing lack of knowledge about a principal subject of geekdom. Kleinrock is indeed considered one of a handful of people who literally and truly invented the Internet. Others were Vint Cerf, Jon Postel, Bob Kahn, and Lawrence Roberts. The Internet didn't come from a vacuum: buy a book, take an hour, and learn about its history.

I think there's growing evidence that the marketplace will soon find its own solution to dawdling on the last mile over monopoly-owned wires.

Wireless.

Already I see where more than a few people are foregoing traditional land-line voice service in favor of cellular wireless phones. That same trend you're seeing for plain old voice traffic will be mimicked for IP service.

With all the war{walking,chalking,driving,flying} going on, I can see where a few strategically-placed public access points (maybe 802.11a with directional antennas) will start an avalanche of users to using wireless for their IP service needs.

Maybe then some of us poor slob end-users can start to see some benefits from that 12 month doubling period for BW/cost ratio on fat pipes.

According to this, there was about 1 gbps of internet traffic in 1995.

If this doubled every 100 days, there would be 50,000 terabits per second of internet traffic today. There's actually less than one terabit/sec of traffic.

By 2010, we could expect more bits per second of internet traffic than there are atoms in the universe.

Chris

Chris

There has been a free version of cscope for quite a while now. It's called cs and available from ftp://cantor.informatik.rwth-aache n.de/pub/unix/.

Both versions work with the graphical tcl/tk interface cbrowser.

--

Anyway the mirror is at http://www.ziplink.net/~ars/slash/rust- box/

I'll delete it in a week. I hope my ISP doesn't kill me - but they should be able to handle the load (T3's etc).

PS. You server might be able to handle the load, but your network pipe seems unable (pages load slowly).