djo7613@blake.acs.washington.edu (Dick O'Connor) (11/28/89)
OK, Ethernet-over-twisted-pair fans, why is this innovation something to get excited over? From theoretical or practical standpoints, why is this new standard a boon to PC networking? I've heard comments about one of the obvious benefits: the cost of twisted pair wiring, as opposed to the (higher) cost of RG-58 thin coaxial cable. Perhaps cabling is easier, too. But is it really cheaper than standard thinnet cable for small groups? Example: There are 7 AT machines in our office that form a workgroup with peripheral and data sharing needs. In addition, there is interest in purchasing "LAN packs" of certain software packages and running them from a server machine. Ethernet cards that support twisted pair run about $100 more than the comparable (WD) standard Ethernet cards (at retail). The cabling *is* cheaper, but this small workgroup doesn't need too much wire (400 feet in a bus). But then you add in the $2500 12-port concentrator necessary (?) for Ethernet over twisted pair, and suddenly you have a *much* more expensive proposition than a standard coaxial bus arrangement. I guess what I'm after is the answer to: Is twisted pair Ethernet a special-purpose solution for large LANs (where wiring costs might predom- inate) or small spurs off a large LAN? Where's the benefit for a small standalone workgroup? "Moby" Dick O'Connor ** DISCLAIMER: It would Washington Department of Fisheries ** surprise me if the Olympia, Washington 98504 ** rest of the Department Internet Mail: djo7613@blake.u.washington.edu ** agreed with any of this!
cliff@violet.berkeley.edu (Cliff Frost) (11/28/89)
I think the biggest loss with pulling thinnet to offices is when you think of the future. What happens when people change offices and someone moves in where you put thinnet. Well, if the new occupant(s) happen to want thinnet then you're fine. If they want something else you're going to have to pull something else. The labor costs quickly dwarf any material costs. Today there isn't much you can't do with twisted pair, including thick ethernet transceiver cable, MacIntoshes and PCs, IBM 3270 type terminals, etc. (I used to assume that FDDI would mean pulling fiber through the walls, but I've heard a rumor that at least one company is planning to try to design gear that uses twisted pair to the workstation for FDDI. If this can be done it'll be worth a bundle, you'd only have to pull the fiber into your phone closets.) In your case, since you're working from a very small perspective, then you may not care about the costs of pulling wires again and again. Cliff Frost Central Computing Services UC Berkeley Re: In article <4645@blake.acs.washington.edu> djo7613@blake.acs.washington.edu (Dick O'Connor) writes: >OK, Ethernet-over-twisted-pair fans, why is this innovation something to >get excited over? From theoretical or practical standpoints, why is >this new standard a boon to PC networking? > ...
wsmith@umn-cs.CS.UMN.EDU (Warren Smith [Randy]) (11/28/89)
In article <4645@blake.acs.washington.edu> djo7613@blake.acs.washington.edu (Dick O'Connor) writes: >OK, Ethernet-over-twisted-pair fans, why is this innovation something to >get excited over? From theoretical or practical standpoints, why is >this new standard a boon to PC networking? > >I've heard comments about one of the obvious benefits: the cost of twisted >pair wiring, as opposed to the (higher) cost of RG-58 thin coaxial cable. >Perhaps cabling is easier, too. But is it really cheaper than standard >thinnet cable for small groups? ... example omitted >I guess what I'm after is the answer to: Is twisted pair Ethernet a >special-purpose solution for large LANs (where wiring costs might predom- >inate) or small spurs off a large LAN? Where's the benefit for a small >standalone workgroup? Here are what I see as the advantages of twisted pair Ethernet: 1) Low cost for cable minimal benefit - labor is usually the major expense in a large installation, equipment is in a smaller one 2) Ability to use existing cable plant in some cases 3) Common wiring type (most of your installer types feel comfortable with it) 4) Small cable - fits in conduit easier (not much different from thinnet tho) 5) Don't have to worry about termination 6) Same wiring can have multiple uses (eg. telephone, Appletalk, RS-232) some won't view this as a feature when people plug things in wrong... And numerous other quibbles about how easy/nice/reliable/stupid/junky/etc the connectors are. The IEEE 802.3T standard is not low cost! This will hopefully change as 802.3T devices are mass produced, and vendors start competing. I have some doubts that it will ever be a super cheap option (comments?). Other non-standard twisted pair products can be alot cheaper. I really don't think there is any significant advantage to using a twisted pair product for the small workgroup as you described (7 stations in close proximity). I wouldn't relate the benefit of twisted pair to the size of the LAN, but rather the wiring requirements. Randy ----- -- Randy Smith wsmith@umn-cs.cs.umn.edu ...!rutgers!umn-cs!wsmith
jgd@rsiatl.UUCP (John G. De Armond) (11/28/89)
In article <4645@blake.acs.washington.edu> djo7613@blake.acs.washington.edu (Dick O'Connor) writes: >OK, Ethernet-over-twisted-pair fans, why is this innovation something to >get excited over? From theoretical or practical standpoints, why is >this new standard a boon to PC networking? this answer is easy. In most modern offices, there is usually spare telephone wire already installed. This is very important in many offices, especially the Class A digs (read EXPENSIVE) where plenum-grade, union installed coax may cost $50 per foot or more. Twisted pair is actually at an electrical disadvantage but as usual, non-technical business considerations often override the slight technical deficiency. John -- John De Armond, WD4OQC | Manual? ... What manual ?!? Radiation Systems, Inc. Atlanta, GA | This is Unix, My son, You emory!rsiatl!jgd **I am the NRA** | just GOTTA Know!!!
kratz@bnrgate.UUCP (Geoff Kratz) (11/28/89)
In article <4645@blake.acs.washington.edu> djo7613@blake.acs.washington.edu (Dick O'Connor) writes: >OK, Ethernet-over-twisted-pair fans, why is this innovation something to >get excited over? From theoretical or practical standpoints, why is >this new standard a boon to PC networking? [ stuff deleted ] >I guess what I'm after is the answer to: Is twisted pair Ethernet a >special-purpose solution for large LANs (where wiring costs might predom- >inate) or small spurs off a large LAN? Where's the benefit for a small >standalone workgroup? To be honest, UTP isn't great for small groups from a cost point of view, since the cost of the hub makes it more expensive than a thin-net solution. Where UTP shines, though, is in large sites or where you are moving machines around (both within buildings and between buildings) *a lot*. UTP gives you a flexibility in large buildings that, quite frankly, cannot be achieved using thick or thin. We know, because we've tried it. We are running a 7 building metro network spread out over the entire city (hooked into a 6 city internet), and we move people around a fair bit. Since we have 4-wire going to every location in every building, all we do is punch a couple of jumpers on a BIXX panel, and voila! Connectivity. And we can do it for about $500/seat now, and it is decreasing as we get more machines (fills up the ports on the hubs). Network management is *cheap* as well. The hubs, bridges and MAUs that we are using are very reliable, and we only have to fix the odd bad punch. Beyond that, we can alter the network topology (ie: turn a workgroup into a subnet) by simply replacing a bridge with a router (with the attendant IP address changes. That's the tough part!). The benefits for a small group would be in scalability if you expect to grow any in the future. You may have a half-dozen machines now, but if you expect any kind of large growth rate (say ending up with 75 to 100 machines in a year), the you might want to look at UTP. Smaller groups that expect to stay small probably can't find many benefits in UTP now, but may if the price of hubs comes down. -- Geoff Kratz Bell-Northern Research, Ltd. Ph: (613) 763-5784 Internet Systems P.O. Box 3511, Station C FAX:(613) 763-3283 Ottawa Ontario Canada K1Y 4H7 BITNET: kratz@bnr.ca I can put my foot in my OWN mouth, thank you!
laubach@hpiag0.IAG.HP.COM (Mark Laubach) (11/29/89)
Well, 1) if you've got trained phone installation people who know how to deal with twisted pair, then they already know how to wire for this new technology and you don't have to retrain them. 2) Bringing back to a data patch panel (AT&T blocks for instance) allow easy customization of the environment, and the ability to change. 3) Having all the wires in one place makes troubleshooting a lot easier. We have all our offices wired here with 4 4-pair modular outlets. Two are reserved for data, and terminate in a patch panel that we control directly, the other two are for voice/pbx work and terminate where our site phone people care about it. In reality, its about two feet to the left of our data panel. We are amazed at how convenient it is for rewiring the system to change configurations, say to install another diskless unix cluster and to hide the server and nodes behind a bridge regardless of were the workstations are located. Also, the amount of real coax that we have is now minimized to a cabinet of starlan hubs and under the computer room floor. Also, the starlan hubs isolate some problems, mostly electrical, that could potentially bring down a whole network if we had used coax. Hope this helps. Mark Laubach Information Architecture Group Hewlett-Packard Company.
johnl@n3dmc.UU.NET (John Limpert) (11/29/89)
In article <693@rsiatl.UUCP> jgd@rsiatl.UUCP (John G. De Armond) writes: >this answer is easy. In most modern offices, there is usually spare telephone >wire already installed. This is very important in many offices, especially >the Class A digs (read EXPENSIVE) where plenum-grade, union installed coax >may cost $50 per foot or more. Twisted pair is actually at an electrical >disadvantage but as usual, non-technical business considerations often >override the slight technical deficiency. The only problem is that the "spare telephone wire" is usually not real twisted pair cable. You end up having to pay someone to run new cable anyway. Is anyone besides me concerned about the RFI/EMI vulnerabilities of twisted pair? I see enough problems with non-existent or poor shielding in current business/consumer electronics equipment. Is your LAN going to collapse every time someone keys up their radio? Is your LAN going to blanket the area with RF garbage? -- John A. Limpert I'm the NRA! Internet: johnl@n3dmc.UU.NET UUCP: uunet!n3dmc!johnl
kwe@bu-cs.BU.EDU (kwe@bu-it.bu.edu (Kent W. England)) (11/30/89)
In article <4645@blake.acs.washington.edu> djo7613@blake.acs.washington.edu (Dick O'Connor) writes: >OK, Ethernet-over-twisted-pair fans, why is this innovation something to >get excited over? From theoretical or practical standpoints, why is >this new standard a boon to PC networking? > One point no one has made, and it really isn't to the point of your question, but to me is the single biggest reason to go with UTP Ethernet, is that this standard, for the first time, allows the user community to exercise the right to choose the media, independently of the network technology and over the objections of the vendors, who used to think that they controlled the media. The work on 802.5 on UTP is evidence of the same trend. We choose UTP for some specific reasons having to do with large scale networking. Of course, UTP has its limits, but for now you can support all major LAN network technologies and every kind of modern phone service on one standard medium. This is a first for us users. We are now in control of the media, not the vendors. We are also rapidly coming to a consensus on how to wire buildings and campuses. The EIA work, mentioned in another thread of articles here, on building wiring is an example of this. I expect this to continue as we move to advanced media, in particular, fiber optic media. 62.5 multimode with ST and FDDI connectors is rapidly becoming the de facto user-defined standard. All other options are being rejected. This is why so many of us are excited by 10BaseT. We are winning the war over the media and bringing it under our control and choice. We choose UTP and star-wired building distribution so we can manage and maintain it efficiently. We make these choices because we know what it takes to scale the networks up, which we are rapidly doing. Since little workgroups in small spaces don't see these scaling problems, it is natural for you to wonder why we who install and manage big networks can get so excited by something that, on the surface, is not particularly exciting technology. We are also winning the interoperable protocol wars, but that is a topic for another thread. :-) Kent England, Boston University
henry@utzoo.uucp (Henry Spencer) (11/30/89)
In article <218@bnrgate.UUCP> kratz@bnr.ca (Geoff Kratz) writes: >... we move people around a fair bit. Since we have >4-wire going to every location in every building... It strikes me that this is the real key: UTP's advantages are mostly a matter of existing UTP wiring networks. Remove those from the equation and it's not such an obvious choice. -- That's not a joke, that's | Henry Spencer at U of Toronto Zoology NASA. -Nick Szabo | uunet!attcan!utzoo!henry henry@zoo.toronto.edu
markf@censor.UUCP (Mark Fabbi) (11/30/89)
In article <1989Nov29.165655.2151@utzoo.uucp>, henry@utzoo.uucp (Henry Spencer) writes: > It strikes me that this is the real key: UTP's advantages are mostly > a matter of existing UTP wiring networks. Remove those from the > equation and it's not such an obvious choice. -- I'd have to disagree. The requirement to pull wire only affects the initial capital cost. The REAL savings of UTP occur with moves/adds and changes. Adding in some extra instalation costs only pushes the cross over point a little father out. (& likely only an extra month or so). My lab network (with >100 nodes) contains a combination of Ethernet, TRN, LocalTalk, RS-232, and some extra voice services. To add a connection, or change from one network to another takes literally less than one minute. UTP to the desk, with a fibre backbone is the logical wiring scheme to install. ======= -- G. Mark Fabbi ...utzoo!censor!markf Associate Director Bell Canada - Engineering Design
pat@hprnd.HP.COM (Pat Thaler) (12/02/89)
johnl@n3dmc.UU.NET (John Limpert) writes: > > Is anyone besides me concerned about the RFI/EMI vulnerabilities > of twisted pair? I see enough problems with non-existent or poor > shielding in current business/consumer electronics equipment. > Is your LAN going to collapse every time someone keys up their > radio? Is your LAN going to blanket the area with RF garbage? > > -- > John A. Limpert I'm the NRA! > Internet: johnl@n3dmc.UU.NET UUCP: uunet!n3dmc!johnl > ---------- When we initially looked at 10 Mbits on twisted pair, this was one of our first concerns. The 10BASE-T committee looked at susceptability including noise measurements on inplace wiring in actual buildings. We also looked at susceptability to crosstalk from other services on twisted pair. RFI and the ability to meet requirements of FCC, VDE, and others was investigated, as was interference with other services on twisted pair. So my answer is: Given good design practice, a 10BASE-T LAN will not "collapse every time someone keys up their radio" and will not "blanket the area with RF garbage." Twisted pair is a balanced media and performs fairly well as far as RFI/EMI. It is limited in length due to attenuation and crosstalk. Given poor design practice, anything can emit. I had a 1200 baud modem at home which interfered with my 2m (~144 MHz) handitalkie if it was plugged in, transmitting or not. Pat Thaler
david@ms.uky.edu (David Herron -- One of the vertebrae) (12/03/89)
As Henry pointed out, UTP has an advantage when there's already that sort of wiring in place. A lot of places were wired with "extra" phone wiring under the assumption that something computerish would be able to make use of it. It turned out to be a bit of a self fulfilling prophecy -- the vendors saw all this extra wiring (and possibly were a source of advice for some people that the "extra" phone wiring would be useful Real Soon Now) and did networking on it. Now, why is a wiring scheme which was designed for the unique problems of analog circuitry going to be useful for digital circuitry? I don't know, someone want to tell me? As I see it -- without the shielding the cabling is going to affect more of the wiring around it and will in turn be affected more by the wiring around it. There will be crosstalk between adjacent wires. There will be more radio interference in the air -- something which people (pregnant women especially) have been up in arms over in recent years. Are the power levels low enough that the signals aren't measurable beyond the cabling? If so then how is the signal strong enough for the components on the network to deal with them? (that is, make out the signal in the noise from other sources (radio stations and the like)...) Heck, how does the power emmissions from the unshielded cable compare to those of the 100,000 watt FM station a couple miles away or the high tension power wiring ? These are the things I'm worried about with it.. If those concerns, and other related concerns, can be answered adequately then I think there's not much room for complaint. In my old job @ ms.uky.edu we used a combination of thin wire and thick wire coax with appropriate repeaters between segments (DEBET's and DEMPR's). There weren't *any* problems, not even the one that the campus communications guy claimed you have with thin coax pulling out of the systems and bringing down the network. In our case when the building was designed an extra set of conduits was run which was intended for some sort of intercom system which was never installed. So we used that for both our serial and ether cabling. In my new job all the wires look like phone wires. But then that's one of the side effects of working at a phone company (AT&T)... :-) -- <- David Herron; an MMDF guy <david@ms.uky.edu> <- ska: David le casse\*' {rutgers,uunet}!ukma!david, david@UKMA.BITNET <- <- New official address: attmail!sparsdev!dsh@attunix.att.com
pat@hprnd.HP.COM (Pat Thaler) (12/06/89)
david@ms.uky.edu (David Herron -- One of the vertebrae) writes: > As Henry pointed out, UTP has an advantage when there's already that > sort of wiring in place. A lot of places were wired with "extra" phone > wiring under the assumption that something computerish would be able > to make use of it. It turned out to be a bit of a self fulfilling Historically, the origional reason for putting in a lot of wire was that multi-line phones required multiple physical lines. I remember when the secretary's phone had a 25 pair bundle feeding it. More sophisticated PBXs came about that did the switching so that the multi- line phone only needed one or two pairs. > prophecy -- the vendors saw all this extra wiring (and possibly were > a source of advice for some people that the "extra" phone wiring > would be useful Real Soon Now) and did networking on it. > > > Now, why is a wiring scheme which was designed for the unique problems > of analog circuitry going to be useful for digital circuitry? I don't > know, someone want to tell me? The line between digital and analog is not so distinct. Any signal that goes over more than a couple of meters of wire has some change to its waveshape. When data and clock are encoded onto the signal and it is sent over 100 m of wire, you need to do some analog processing (eg filtering, wave shaping, clock recovery) to recover the digital data. You analyze the characteristics of the wire and from that you determine if what you want is feasible and what kind of processing you do. In the case of 10BASE-T, there were some givens: we wanted to use the same encoding and the same encode/decode IC's as existing 802.3 10Mbit implementations. > > As I see it -- without the shielding the cabling is going to affect > more of the wiring around it and will in turn be affected more by the > wiring around it. There will be crosstalk between adjacent wires. Yes, there is crosstalk to adjacent wires. The amount is controlled by filtering. The crosstalk is one of the primary reasons why the distance is limited to 100 m. > There will be more radio interference in the air -- something which > people (pregnant women especially) have been up in arms over in recent years. RFI from 10BASE-T is no more than from 10BASE2. At any distance from the pair, fields caused by the two wires cancel and emmissions are negligable. > > Are the power levels low enough that the signals aren't measurable > beyond the cabling? If so then how is the signal strong enough for the > components on the network to deal with them? (that is, make out the > signal in the noise from other sources (radio stations and the like)...) Because, the receiving device is looking at the differential signal sent over a balanced media. If the media was perfectly balanced and the transmitter emitted no common mode signal onto the media, there would be no emmissions regardless of signal level. Similarly for susceptablity. Of course, balance is not quite perfect, so we do things to limit emmissions and susceptability. We limit the signal strength, filter the transmitted signal at about 15 MHz, filter the received signal to remove most of the noise, implement a squelch, etc. > Heck, how does the power emmissions from the unshielded cable compare > to those of the 100,000 watt FM station a couple miles away or the > high tension power wiring ? > Emmissions from properly designed 10BASE-T equipment (including the attached wiring) can meet regulatory requirements and are comperable to that of any other modern computer equipment. That means that they are a lot lower than emmissions from older equipment since the FCC regulatory requirements are fairly recent. > > These are the things I'm worried about with it.. If those concerns, > and other related concerns, can be answered adequately then I think > there's not much room for complaint. > > > In my old job @ ms.uky.edu we used a combination of thin wire and thick > wire coax with appropriate repeaters between segments (DEBET's > and DEMPR's). There weren't *any* problems, not even the one that > the campus communications guy claimed you have with thin coax pulling > out of the systems and bringing down the network. > > In our case when the building was designed an extra set of conduits > was run which was intended for some sort of intercom system which > was never installed. So we used that for both our serial and ether > cabling. > > In my new job all the wires look like phone wires. But then that's > one of the side effects of working at a phone company (AT&T)... :-) > -- > <- David Herron; an MMDF guy <david@ms.uky.edu> > <- ska: David le casse\*' {rutgers,uunet}!ukma!david, david@UKMA.BITNET > <- > <- New official address: attmail!sparsdev!dsh@attunix.att.com > ----------