mats@oblio.UUCP (Mats Wichmann) (08/08/89)
We are looking for experiences / impressions of running twisted-pair ethernet. Last time I looked at this stuff it was all in the talking stage - nobody had any commercial hardware. Now I understand Cabletron makes some equipment, and there are presumably others. I wasn't paying any attention before, but now we about to move and looking at cabling options for cubicles, especially for use with our X terminals. I have a notion this has been discussed here before and I missed it (hey, I don't have time to read everything); if so, I think I would appreciate mail rather than followups. Mats Wichmann Acer America hplabs!oblio!mats
craig@hprnd.HP.COM (Craig Blackwood) (08/09/89)
> We are looking for experiences / impressions of running twisted-pair > ethernet. I think it is great! The diskless workstation from which I am writing this is connected to the server via Twisted Pair ethernet (HP StarLAN 10). With other types of cable (thick and thin coax) one person can mess up the whole network simply by messing with the cable. With StarLAN 10 other users can do whatever they want to the twisted pair that runs to their machine without causing any problems for others. > Last time I looked at this stuff it was all in the talking stage - > nobody had any commercial hardware. The product is a reality. The standard is almost a reality. The IEEE standard is called 10baseT and (from what I know :-) will be official near the beginning of next year. > Now I understand Cabletron makes some > equipment, and there are presumably others. I wasn't paying any attention > before, but now we about to move and looking at cabling options for cubicles, > especially for use with our X terminals. HP makes twisted pair ethernet products (called StarLAN 10) including PC LAN cards, HUBS, minicomputer and Workstation interface cards. AT&T and Synoptics also make StarLAN 10 products. There are probably many others. > > I have a notion this has been discussed here before and I missed it (hey, I > don't have time to read everything); if so, I think I would appreciate mail > rather than followups. > > Mats Wichmann > Acer America > hplabs!oblio!mats > ---------- Craig Blackwood craig@hprnd.rose.hp.com
mkd@mtunh.ATT.COM (Mark Darby) (08/10/89)
In article <2230020@hprnd.HP.COM>, craig@hprnd.HP.COM (Craig Blackwood) writes: > > We are looking for experiences / impressions of running twisted-pair > > ethernet. > > I think it is great! The diskless workstation from which I am writing this > is connected to the server via Twisted Pair ethernet (HP StarLAN 10). With > other types of cable (thick and thin coax) one person can mess up the whole > network simply by messing with the cable. With StarLAN 10 other users can > do whatever they want to the twisted pair that runs to their machine without > causing any problems for others. This is true for a standalone TPE (twisted pair ethernet) network, but with the proper equipment, one can expand an already existing Ethernet/ Cheapernet network with StarLAN 10, providing cost effective expansion without losing the original network investment. As a footnote, AT&T's TPE solution is also called "StarLAN 10". > > > Last time I looked at this stuff it was all in the talking stage - > > nobody had any commercial hardware. > > The product is a reality. The standard is almost a reality. The IEEE standard > is called 10baseT and (from what I know :-) will be official near the beginning > of next year. > 10BASE-T is a task force under the auspices of IEEE 802.3, drafting a standard for 10Mb/s baseband transmission over twisted pair wire. The 10BASE-T draft was recently voted upon by IEEE 802.3 voting members and failed by a significant margin (75% ratio of yes votes to yes+no votes is required for approval, actual ratio obtained was about 59%). Over 600 comments were received from voters which must be addressed, however most of these were editorial comments and general specmanship issues. As a matter of fact, the 10BASE-T task force is meeting this entire week in San Jose, California to address voter comments and issues. The new schedule of target dates for 10BASE-T includes a second letter ballot for the end of this year, with the presenting of voting results to IEEE 802.3 by March 1990. Assuming no delays in the entire schedule, including submitting draft to the IEEE Computer Society and IEEE review board, the draft could be sent to ANSI for submission to ISO by 4Q90. This assumes all submissions will result in approvals every step of the way. Any disapproval along the way could delay the progress of the draft further. > > Now I understand Cabletron makes some > > equipment, and there are presumably others. I wasn't paying any attention > > before, but now we about to move and looking at cabling options for cubicles, > > especially for use with our X terminals. > > HP makes twisted pair ethernet products (called StarLAN 10) including PC LAN > cards, HUBS, minicomputer and Workstation interface cards. > AT&T and Synoptics also make StarLAN 10 products. There are probably many > others. AT&T, HP and Ungermann-Bass currently make TPE hardware products which comply with key technical parameters of the 10BASE-T draft, including transmit voltage levels, transmit equalization, 100m distance support per twisted pair link, and the use of a multiport repeater (ala IEEE 802.3) as the HUB element. Because of this, the current hardware from these vendors INTEROPERATE, which is what standards are all about. SynOptics' TPE solution is called "Lattisnet." They have made some new product announcements, and I do not know if their new products take any more 10BASE-T parameters into consideration. I do know that previous products (with some exceptions) from Lattisnet took very few key 10BASE-T parameters into consideration. As the 10BASE-T draft becomes more a reality, each vendor will have a migration path from current product to final "10BASE-T" product. Those vendors who have proprietary solutions to TPE will a more difficult migration path to follow, depending upon how much the vendor has veered from the path followed by 10BASE-T. > > > > I have a notion this has been discussed here before and I missed it (hey, I > > don't have time to read everything); if so, I think I would appreciate mail > > rather than followups. > > > > Mats Wichmann > > Acer America > > hplabs!oblio!mats > > ---------- > > Craig Blackwood > craig@hprnd.rose.hp.com Let me know if I can be of further assistance. ------------------------------------------------------------------ Mark K. Darby AT&T Bell Laboratories AT&T: (201)957-2706 200 Laurel Ave. uucp:..!att!mtunh!mkd Middletown, NJ 07748
djh@cci632.UUCP (Daniel J. Hazekamp) (08/11/89)
We recently installed a David Systems ExpressNet twisted pair ethernet hub. The hub has 12 TPE ports, and 1 AUI port for connection to an E-net backbone. The hub is a 10BASE-T compliant (we all know the 10BASE-T story by now) multiport repeater, and successfully interoperates with AT&T STARLAN 10. The reason we went with David Systems rather than AT&T, is that the David hub worked successfully over our installed 'shielded' RS232 cabling. -- Dan Hazekamp rochester!cci632!djh Computer Consoles Inc. (CCI) uunet!ccicpg!cci632!djh Rochester, NY uunet!rlgvax!cci632!djh Internet: cci632!djh@cs.rochester.edu
pat@hprnd.HP.COM (Pat Thaler) (08/12/89)
Mark Darby writes: > > The new schedule of target dates for 10BASE-T includes a second letter ballot > for the end of this year, with the presenting of voting results to IEEE 802.3 > by March 1990. Assuming no delays in the entire schedule, including > submitting draft to the IEEE Computer Society and IEEE review board, the > draft could be sent to ANSI for submission to ISO by 4Q90. This assumes > all submissions will result in approvals every step of the way. > Any disapproval along the way could delay the progress of the draft further. > Actually, the draft goes to the Technical Committee on Computer Communications (TCCC) and the IEEE Standards board after 802.3 approval. Depending on whether it needs a full reballot or only a reconfirmation ballot in 802.3 it could get approval as an IEEE/ANSI standard sometime between 2Q and 4Q90. As I explained in a previous posting, NO voters have the opportunity to change their votes as we respond to the comments on the ballot. Therefore, the result of the ballot can change. Assuming that we satisfy the NO voters, a reconfirmation ballot gives all the voters on the origional ballot a chance to see any changes we made in resolving the ballot and to confirm that they still find the draft acceptable. The balloting of a draft is a concensus process, not a simple 75% passes and less than that fails. Efforts are made to resolve all negative votes. I expect the draft to be submitted to ISO to start the balloting process there at the same time the TCCC ballot is conducted. I'm not sure how long it will take to complete ISO balloting. Pat Thaler Opinions expressed are my own and not necessarily those of the 10BASE-T task force.
ncpjmw@amdcad.AMD.COM (Mike Wincn) (08/14/89)
In article <692@mtunh.ATT.COM> mkd@mtunh.ATT.COM (Mark Darby) writes: >10BASE-T is a task force under the auspices of IEEE 802.3, drafting a >standard for 10Mb/s baseband transmission over twisted pair wire. The >10BASE-T draft was recently voted upon by IEEE 802.3 voting members and >failed by a significant margin (75% ratio of yes votes to yes+no votes is >required for approval, actual ratio obtained was about 59%). Over 600 >comments were received from voters which must be addressed, however >most of these were editorial comments and general specmanship issues. Your description makes it sound as if 802.3 first approved and then rescinded the 10BASE-T draft, when what actually happened was that it was overwhelmingly APPROVED for letter ballot, and ballot comments came in at 59% as you point out. It should be mentioned that the ballot process allows three possible responses: Approve, Approve with Comment, and Disapprove, and it is the ratio of disapproves to approves that count. Further, many of the disapproves were of an editorial nature - that is, the voter's comments had nothing to do with feasability or functionality, but with description instead. An example that comes to mind is 'differential input voltage' rather than 'differential voltage' or '...the output voltage is 2.2v peak' rather than '.. peak voltage is 2.2v'. It is believed that many of the 'disapprove' comments will be converted to 'approve' with an editorial pass. It should also be pointed out that many of the comments were duplicates, and that while resolving all comments is not a trivial task, it will not be nearly as difficult as an attempt to resolve 600 _unique_ responses. Another point you failed to mention is that we've not seen any 'show stoppers' - i.e. issues that would require complete rewrite/reballot - and it is thus presumed that the task remaining is well-defined and 'short term'. An implication of this is that end users run _lower_ risk as a date to confirmation ballot draws near, since OEMs will have already incorporated the most recent updates in their products. >The new schedule of target dates for 10BASE-T includes a second letter ballot ^^^^^^^^^^^^^ Actually, the goal is a 'confirmation ballot', which is very different. The ability to realize that goal depends on how much progress the Task Force can make before November. >------------------------------------------------------------------ >Mark K. Darby Mike Wincn ncpjmw@amdcad.AMD.COM (408) 749-3156
hagan@hagan.dccs.upenn.edu (John Dotts Hagan) (08/30/89)
I have had quite a few debates and discussion with vendors, and other people responsible for campus networks, about the use of twisted pair ethernet vs. Thinwire ethernet (RG58 coax). I have always favored the coax, because we always seem to have to install new wire to support every new station, since we can never find where old (pre our group wiring the campus) copper wire goes. The cost to install thinwire vs. twisted pair is not large, since the bulk of our cost is the labor to install. Our standard wiring is to pull tree kinds of wire to all stations: RG58 coax, and both shielded and unshielded 22 guage twisted pair. The shielded wire is up to IBM's 16 Megabit specs, but we have never used the IBM token ring over that wire. Sometimes we use the shielded for a second asynch connection, although the shield causes some problems over long runs. We also pull telco wire, if we are in a new building and our telco group works with us. We have also planed for video, but never installed it. Finally, whenever we put in conduit for the wire, we make sure it could hold fiber (no bad bends). Anyway, I believe the twisted pair solutions are near the capacity of twisted pair, and some products have distance limits that are close to some of our longer station to closet runs. However, RG58 has no trouble reaching any desk. My final trump card is I hope that some smart vendor makes a faster version of ethernet (or even FDDI) that can run over RG58 coax. I think anything in the range of 100 Megabits is next to impossible for the twisted pair in our walls, but should be quite possible over the coax. So maybe we won't need to rewire for fiber to all desks to get at least FDDI-like speeds. --Kid.