djo7613@blake.acs.washington.edu (Dick O'Connor) (11/30/89)
Thanks to all of you who responded, both here and via email, to my questions about using Ethernet over twisted pair. The consensus seems to be that it is the wiring medium of choice where wiring installation will be the biggest expense. Other solutions probably are more optimal for small, isolated workgroups like the 7-station group I used as an example. Here are the comments I've received via email to date, excerpted (in *some* cases, at least :) with permission... ******************************************************** >From: wsmith@umn-cs.CS.UMN.EDU (Warren Smith [Randy]) 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 ----- ******************************************************************* From: osterman@snmp.ocf.llnl.gov This is an exciting new boon to the vendors of Ethernet equipment. Just imagine, now anyone who has a phone has the capability to install the vendor's Ethernet equipment. The greatest cost of any wire is the installation cost. If you have to install twisted pair then the cost of installing RG58 is not an inmportant cost difference. More can be done with the twisted pair however. (RS232, Apple Localtalk, etc.) There was talk from some FDDI vendors about CDDI (Copper Distributed Data Interconnect) using RG58, but I haven't heard anything about it in quite a while. >Perhaps cabling is easier, too. But is it really cheaper than standard >thinnet cable for small groups? Cabling is harder because you have more wires to deal with. (I had forgotten what a pain twisted pair was until I had to try to figure out why a particular Ethernet installation wasn't working and found that I had all of these possible wire combinations to deal with. (This installation used 3 pairs out of 4 and they had to be intermixed to satisfy both DEC and AT&T. The installers did the most logical thing and wired it one for one.) It is not cheaper to use twisted pair if you have to install wire, unless you worry about pennies per foot. >...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. That $2500 concentrator may be the best thing that you ever put on your network. If it includes management and control capabilities you'll love it as the network grows. >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? All LANs start as small workgroup LANs. The trouble is that if you design them as small workgroup LANs you often have to redesign them later. If you design them as part of a large network at the beginning then they work just fine in the future and if the network never gets any larger, well, it's only money. (How much does it cost if your network goes down and the people who use are unproductive until it comes back up? $50 to $100 per hour per person?) One last thing. If thin wire has multiple taps on it then you better be able to see it all from one place or you'll regret it. ********************************************************** From: "Steven M. Miller" <smiller@umn-cs.cs.umn.edu> For small groups that don't move much it's no big deal. But, for medium and large size groups or small grups that move around a lot it's ideal. To hook my Sun up in a different office I'll just have to go to a patch panel and make 1 wire move and my Sun is hooked up if I'm using twisted pair. The initial costs are higher to install, but the costs of moving people around and adding new workstations are significantly lower. Plus I don't get the ugliess of thick or thin net cables. -Steve ************************************************************* From: Ken Yap <ken@cs.rochester.edu> I don't know the exact economics of this but my friend who works for a NYC brokerage firm was saying that twisted pair is a candidate where the wiring is already in place or the conduit space is limited, e.g. old buildings, and pulling thick ethernet cable through is a big hassle. I believe there is a shorter distance limit on thin ethernet, no? It looks like for new installations there is no advantage over thin ethernet as in your case. ************************************************************** From: bob@aecom.yu.edu (Bob Lummis) I doubt if anyone would claim ethernet over twisted pair would be better/ cheaper for connecting a small work group. The reason people get excited about it is that if your task is to provide a network for a whole campus it allows you to use twisted pairs that are already in the walls rather than pulling new cable. In a campus-wide network installation project the labor costs of installing cable are the most expensive item in the whole project. --- Robert C. Lummis, Director of Scientific Computing Albert Einstein College of Medicine, Bronx, N.Y. 10461 bob@aecom.yu.edu ...!uunet!aecom!bob *************************************************************** "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!
sward@cfa.HARVARD.EDU (Steve Ward, CF) (12/01/89)
There is a major reason we find twisted pair ethernet attractive at our site: network bandwidth management and conservation in the face of growing numbers of remote file server/multiple client workstation groups. We already have multiple local backbone cables with departments isolated behind bridges. However, callaboration spreads across departments and divisions, so that we have a growing trend of computer workstation clusters (be it VMS VAXclusters or Unix/Ultrix/NFS distributed file system clusters) where the file server and the client workstations are physically spread between all departments, divisions and possible geographic locations, and mobile on top of it all! If the goal is to keep all such file server clusters isolated on their own ethernet segment behind a bridge then it is nigh impossible under the conditions I describe. Enter twisted pair: Aha! Now each node, no matter its location, get connected to a central wiring hub. At wiring hub clusters of machines, independent of physical location, can be placed on the same segment which can be bridge isolated. Again, small sites may be able to easily deal with all circumstances using thinwire or thickwire. We have hundreds of workstations and must support the network connection of many such server/client clusters of workstations. Twisted pair provides a method to give our users what they need in terms of connectivity and flexibility while allowing us to keep local server/client traffic off the backbones. As a plus, the bridges, repeaters, etc. are automatically located in our central equipment area where we have our wiring center. We have always endeavored to locate all network equipment here but it has not always been possible. We are eager to see the further development and maturation of this network technology. Like everyone else we are waiting for standards, both official and marketplace ad hoc. In the meantime we are testing and evaluating. (seemingly the almost-fulltime fate of every network manager and planner). Steven M. Ward ward@cfa.harvard.edu
wsmith@umn-cs.CS.UMN.EDU (Warren Smith [Randy]) (12/01/89)
In article <249@cfa.HARVARD.EDU> sward@cfa.HARVARD.EDU (Steve Ward, CF) writes: >There is a major reason we find twisted pair ethernet attractive at our >site: network bandwidth management and conservation in the face of >growing numbers of remote file server/multiple client workstation >groups. [text deleted] >..... Enter twisted pair: Aha! Now each node, no matter its >location, get connected to a central wiring hub. At wiring hub clusters >of machines, independent of physical location, can be placed on the same >segment which can be bridge isolated. At the risk of beating this subject to death, I would like to point out that you can use the exact same configuration using thinnet. Run one thinnet to each device and connect them all to standard thinnet multiport repeaters. There are even some advantages to doing it this way: 1) longer distance per run 2) lower cost - the twisted pair equipment is alot more expensive when your devices have built in thinnet xcvrs (as most now do) You can but an 8 port thinnet multiport for <$2K and 4 port <$1.4K 3) more noise immunity/less RF 4) you can daisy chain off another device if a quickie install is needed Of course, there are disadvantages to thinnet as well. Two that are relevant to the star topology are: 1) thinnet multiports are rather stupid about disconnecting screwed up segments, which can result in a hung network. 2) thinnet patch panels/closet hardware/wall plates/etc are harder to come by and more expensive than telco type stuff. Randy -- Randy Smith wsmith@umn-cs.cs.umn.edu ...!rutgers!umn-cs!wsmith
illgen@hq.af.mil (Keneth..Illgen) (12/01/89)
What I find, typically, as one who provides different media-services
to a wide range of applications is that most of those who are tied into
a twisted pair network immediatly question the future need for increased
bandwidth. At this point they start looking at various ETHERNET forms.
In most buildings twisted pair is already there with only the occasional
jumper needed. Even if it isn't installed it is easily the simplest to
install. The attractiveness to me is that it takes an act of God to kink the
stuff so the need to replace a good shielded twisted pair is rare after the
initial installation.
On the down-side is noise susceptibility. Fortunatly for most of you
there isn't the RF floating through your buildings as we here in the Pentagon
experience; but it can weigh heavily depending on where you run the cable and
how well the cable is shielded. Also twisted pair is easy to tap; if some-
one so desires.
My overall 2 cents on twisted pair is that it's overrated. The pluses
are in cost savings and installation. That makes it easier for a network
planner to go with it (especially if it is already installed). But as a
network grows the data transfer reliability will drop, the noise suscept-
ibility will get higher, security will be harder to maintain and you can't
run long legs; even if you fan-out.
As an alternative I would suggest fiber. Of course that isn't always
agreeable with the budget. So the next best thing would be to run a riser
with thick E-NET and legs of either thick E-NET or thin net. Thick E-NET is
not difficult to install. As one who has done more then my share of E-NETs
I can tell you that the primary concern of a data cable installer is not the
weight of the cable but whether or not it's going to kink up going
around corners and through walls. A small cable like TWINAX is a lot harder
to run then thick E-NET (IMO). I also feel that people shouldn't be so hard
on thinnet. Granted adding taps can be a pain but if the run is planned right
'T' connectors can be installed at points where cables can be moved or
eliminated with little difficulty.
"Where there's a will...There's an 'A'" - John Ritter
Ken Illgen
HQ USAF LAN
The Pentagon
Washington, DC
Phone: (202) 695-5090 Mail: illgen@hq.af.mil
Autovon: 224-6136pat@hprnd.HP.COM (Pat Thaler) (12/06/89)
> > At the risk of beating this subject to death, I would like to point out > that you can use the exact same configuration using thinnet. Run one > thinnet to each device and connect them all to standard thinnet multiport > repeaters. There are even some advantages to doing it this way: > > 1) longer distance per run > 2) lower cost - the twisted pair equipment is alot more expensive > when your devices have built in thinnet xcvrs (as most now do) > You can but an 8 port thinnet multiport for <$2K and 4 port <$1.4K You can buy a 12 port twisted pair hub for <$2K. More devices with built-in twisted-pair ports are becoming available. As integrated transceiver and repeater chips become available, I expect prices to drop. > 3) more noise immunity/less RF Bit error rate and RF is about the same for the two media. The distance of twisted-pair was limited to 100 m in order to attain the 1 in 10**9 BER. Of course if you are in a very noisy area, you may be better off with fiber than with either electrical media. Another option is to run over twisted-pair cable with an overall shield. This is a truely shielded cable as opposed to coax where the shield is used for signal return path. > 4) you can daisy chain off another device if a quickie install is needed > > Of course, there are disadvantages to thinnet as well. Two that are > relevant to the star topology are: > > 1) thinnet multiports are rather stupid about disconnecting screwed up > segments, which can result in a hung network. > 2) thinnet patch panels/closet hardware/wall plates/etc are harder to come > by and more expensive than telco type stuff. > > Randy > -- > Randy Smith > wsmith@umn-cs.cs.umn.edu > ...!rutgers!umn-cs!wsmith > ---------- There are certainly useful applications for thinnet because of its greater distance and its ability to daisy-chain. Twisted-pair with a fiber or coax backbone works well in a lot of office environments. I have worked on the development of both 10BASE-T and 10BASE2. They are not mutually exclusive, some LANs should use both. Pat Thaler
karn@jupiter..bellcore.com (Phil R. Karn) (12/07/89)
On the subject of Ethernet and RF interference, I can say with some authority that coax is *not* a guarantee against problems. I have a thin Ethernet at home with three machines (Sun, XT and 386 PC clones). When I replaced my PCs' old decrepit 3Com 3C-501 Ethernet controllers with newer TRW PC-2000 boards, the level of RF noise heard in my HF (< 30 MHz) amateur radio transceiver went up significantly. It seems that as competition heated up in the PC Ethernet market, the manufacturers started cutting corners on their RF filtering. The root problem is the requirement that Ethernet coax shields be isolated from the host computer ground. Ethernet controllers with built-in thinnet transceivers do this with an insulated BNC connector on the back panel. Unfortunately, the isolation between the transceiver and the rest of the unit is not very good at RF frequencies (probably due to capacitive coupling across the isolation transformers, plus other stray board capacitances). The end result is that the coax acts as a nice long-wire antenna that effectively conducts RF noise out of the computer and radiates it. Note that the coax shield is ineffective here because the noise is a common-mode signal; radiation is from the outside of the shield. The very early 3C501 cards had reasonably effective pi-network low pass filters on their BNC connectors that consisted of ferrite rings plus a couple of disk capacitors from each side to the metal mounting bracket. Later versions of the 3C501, plus just about every other manufacturers' cards, started scrimping on the filters. Some current cards have no filtering on the connector at all, but instead have a single disk capacitor/spark gap on the PC board between the transceiver and PC grounds. This probably *enhances* RF leakage because of the "hot" internal grounds found in most computers. (Many digital engineers have apparently never heard of "ground loops".) I finally got my problem under control by using external transceivers. Ethernet transceiver cables are well shielded, and more importantly their shields connect directly to the external metal casing of the PC and to the metal chassis of the transceiver. Also, the capacitance of the cable apparently does a good job at attenuating conducted noise between the signal pairs and the cable shield. It's still not perfect, but about the only better thing I could try would be fiber. By the way, for a good time try monitoring an Ethernet with a shortwave receiver tuned to 10 MHz. File transfers sound remarkably like the infamous Russian Woodpecker (a Soviet over-the-horizon radar that is much reviled by radio amateurs and other users of the HF spectrum because of the widespread interference it causes.) Phil