kiessig (12/10/82)
With so many sites and so many users and such LARGE phone bills (especially for places like decvax) (have you ever considered that USENET is really a vast money-making plot by AT&T?), there has got to be a better way. I have two ideas: To reduce overall traffic (less redundancy) and to increase usefullness, I agree that we need higher throughput. This can be done in a couple of different ways: 1. Get out from under the phone company (I can hear the missles coming...) With more and more sites joining the net, we could together almost afford to put up our own satellite (call it UUSAT?). Barring that, a more reasonable approach would be for the high traffic centers on each side of the country to rent time on a satellite channel. A small 3 or 4 foot dish should do the trick. Sites would be given "broadcast windows" during which all other subscribers would listen and pick up the latest news (don't TV networks already do something like this?). This news could then be forwarded fairly quickly to neighboring sites. I honestly believe the overall cost for such a network would be MUCH LESS than the current system. 2. Increase connectivity By reducing the maximum time from one end of the netowork to the other we would get better service, but not lower cost. We could do this by tying some of the hubs together better, and finding sites to volunteer as new hubs (our site would consider such a proposal, for example). This might also be done with leased lines in an attempt to cut costs. I also think it would be a good idea to establish a "network fund" to which usenet sites would be asked to contribute. This fund would then be distributed to non-Bell sites with large phone bills to help offset costs of things like leased lines and satellite time. Contributions would not be mandatory, but rather on a "as you can afford" basis, with some suggested minimum. 400 sites X $50/month is $20,000/month - that could go a long way toward making this network better - we might even be able to hire a programmer or two to rewrite uucico.... Rick Kiessig {sri-unix,megatest,dsd,randvax}!fortune!kiessig
swatt (12/10/82)
I think it was at the Santa Monica USENIX conference someone suggested using HAM radios to do the long-haul transport (say between decvax and ucbvax). I'm told it's now legal to broadcast ASCII on HAM frequencies and the effective rate is 600baud. For a permanent connection, I think this rate would suffice. Once the equipment is purchased, your transmission costs are power and maintenance. I know zilch about HAM, but the idea struck me as one well worth pursuing. Anybody out there comment on feasiblity/cost? What you would want to do in such a situation is make the uucico daemon permanently attached to a given line. I've looked into that and it's not difficult. Simply instead of acknowledging the HANGUP request, you wait 30 seconds or so (depending on the packet timeout) and respond 'H' 'N' (Don't hangup), switch roles, and go through the cycle again. Every 30 seconds the two would swap roles; the MASTER system would scan for work (you would want to make this a bit smarter too), and everything else would go on as normal. We will probably end up doing exactly that when "wexelvax" comes up in Shelton and we have a permanent 9600baud leased line connecting the two systems. - Alan S. Watt PS You could still stay with AT&T and still get more bits for the buck: install leased lines between the major nodes. This has the problem of splitting costs between two organizations, plus it freezes the current connectivity of the net, which tends to change too rapidly.
sjb (12/11/82)
Using amateur radio to carry the bulk of the network traffic is certainly interesting, though I'm not sure it can be done. For one thing, there would have to be some research done into just what kind of thing we'd be getting into. Most noticeably, amateur radio third party traffic (i.e. traffic handled by individuals on behalf of others) rules and regs are probably the most complex and complicated of all the ham rules. Though logging requirements are being eased, I still believe amateur stations are required to keep copies of all third party traffic handled for one year -- this alone would probably do the suggestion in. Automatic control of amateur radio stations, while allowed, is another sticky area to get into. Station identification (required at ten minutes intervals among other times) would increase transmission time. An exciting proposal, but perhaps not very feasible.
jwb (12/12/82)
Amateur radio for net traffic has other problems. Keeping the thing on the air all the time is difficult. Spectrum space is more precious (in real terms) than phone lines. Also, unless I am mistaken, a control operator has to be truly in control at all times (although remote control is OK under cer- tain stringent circumstances). This has been waved for experimental beacons, etc., but I think this would be unlikely for third party traffic (which had a second class status to the regulators anyway). Jack Buchanan K4FRS
bstempleton (12/13/82)
Ham is a good idea, but taking a dedicated frequency is not a good idea - I suspect other ham people might complain. At any rate, who says uucico is the best program to use - surely we can come up with something not as grossly inefficient on large bunches of files. I suspect that a better idea than HAM is public networks like telenet/tymnet/datapac. These are all interconnected, and will provide 1200 baud to just about anybody and far more if you pay them a bit of money. With the amount of money spent on usenet, a bunch of sites together could easily buy a large 68000 system or perhaps pdp-11 and put it on one of the networks. If they did this (you might need a slightly different program) you could have a machine supporting 20 or more sites, and saving each of them a bundle. These machines have the advantage that anybody in any major city call call them locally, and you only pay about $6 per hour. Reducing the decvax<->ucbvax link would be nice, but as far as I know news crosses the continent in several places. I know that decvax in NH feeds microsoft in WA a few other west coast people. For news, at least, there should be only one cross-continent gateway. The only trouble with all this is that we have no central authority, and if you actually had to go to your boss to say "Can I have $2000 to by a 68000 unix bos to run the net with 20 other sites" rather than having him/her sign the phone bill cheque every month, you would run into trouble. Brad
karn (12/14/82)
Regarding Alan Watt's comments about using amateur radio (ham radio) for usenet transmission: There is quite a collection of us hams who are now actively working on equipment and protocols for packet data transmission. In fact, planning for use of frequencies in the new amateur satellite Phase 3-B (to be launched 4/83) includes a dedicated packet radio channel. I would like nothing better than to see a new crop of people come into amateur radio to develop the potential of this technology. However, there are some VERY IMPORTANT legal problems. 1. By law, the amateur service is forbidden both to charge for its services, or to be involved in any "pecuniary" (moneymaking) activity. However, we justify Usenet to our companies by its aiding us in the work we do. 2. Although the rules regarding digital transmissions have recently been greatly relaxed (there are now very reasonable limits on bandwidths on the VHF bands, and essentially no limit on frequencies from 1215 mhz up) the FCC is still somewhat touchy regarding what they call "third party traffic" (relaying traffic originated by non-hams), particularly if it is automated. Their concern stems from the ease at which this kind of service can be used for business activities. The rules require that automatic phone patches on FM repeaters, for example, record all conversations and make them available on demand to FCC engineers. The equivalent requirement here would be to keep all netnews articles available for FCC perusal, and the operators of all stations involved would be responsible for any abuses. I'm not exactly sure I'd want to risk my license on this net. On other words, using amateur packet radio for usenet transmission is becoming technically feasible, but not legally so. Phil Karn, KA9Q/2 BTL Murray Hill
stevenm (12/15/82)
Perhaps the correct thing to do at this stage is to form a non-profit corporation, USENET, Inc., to administer the network, setup the suggested news-server nodes, and so forth. I suspect that there might be a bootstrap problem - acquiring sufficient funds to start might be difficult. People are always a little leery of paying for what they once got "free" (never mind that it really wasn't free). I think that this corporation would need somewhere in the vicinity of $100,000 to start up in a big way. One possible approach is to make USENET a bona fide branch of USENIX, Inc. It is likely that every site on the net is a member of USENIX (except, perhaps, the Bell Sites). The start-up money could come from a surcharge to the membership fees charged to USENIX members. Comments? As always, address them to me and I will summarize. S. McGeady (decvax,ucbvax)!teklabs!stevenm
dave (12/16/82)
I agree with the telenet/tymnet/datapac suggestion. In addition, there is Dataroute in Canada and probably something similar in the U.S., for large-volume point-to-point connection (e.g., U of Toronto to U of Waterloo, etc.?). Dave Sherman
swatt (12/17/82)
Regarding Telenet and friends: Last time we talked to them, they were EXPENSIVE (no I don't recall the figures). They also couldn't get us to Europe so we simply dropped that idea. Now perhaps we didn't ask the right questions, which are: 1) To dial OUT via Telenet, is it necessary to have one of their boxes installed in your computer room, or can you just dial out to a local number using PSN and pay Telenet on a connection+packet basis? 2) To receive calls routed through Telenet, is it necessary to have one of their boxes installed in your computer room, or can you have calls routed through the PSN from your local Telenet office? Their representative told us the Telenet assumptions are: 1) You have a computer to which a bunch of people want access. 2) Telenet installs equipment in your machine room and runs a leased line or three to their nearest office. 3) People attach modems to terminals, call up their local Telenet exchange, and enter whatever magic code numbers signify your location. They get a connection (300 or 1200 baud), with guaranteed data transparency and integrity. To them it looks like an ordinary dialup with some funny business at the front; to you it looks like an ordinary dialup. In other words, when the system was originally put together, they didn't consider the case of a computer calling another computer as the typical one. As I recall, most of the cost was associated with installing their equipment in your machine room; the connection time and packet charges seemed quite reasonable. We also talked to Comsat, and their approach is to rent you time on a channel of up to 250Kbits/sec, but you need a satellite antenna on the roof (leased figuring cost $150K @ 3%/month). At that time they couldn't get us to Europe either (since changed; they've signed an agreement with the U.K. Post Office). Could somebody out there who uses Telenet or similar service post the actual usage requirements, costs, etc., and how sites would go about making use of it for USENET? - Alan S. Watt