'I wonder what this cable does': How to tell thicknet from a thickhead Is a Loose Cannon worse than a Big Cheese? What happens when the two are combined? Stir in some overconfidence and you have today's entry in the On Call archives. Our tale today comes from a reader Regomized as "Jon", who was working for a small software development company in Cupertino, California (no, not that one.) One of …
You used to be able to buy a backpack which fitted an original mac plus. With space for keyboard and mouse. It was rather blocky. I used to have a pack for my motorbike like that. Pillow fitting which slid over the luggage upright. Rudimentary straps. On long trips I would turn it around to sit on the back seat meaning I had a back support.
Back in the day when ISDN was popular I was tasked with putting some audiovisual kit in one of a bank's main offices in Northants.
On the top floor the C-levels sat along with the videoconferencing room.
To be fair the kit was pretty impressive - it was one of the early cameras that could track a person as they moved around a room and it all went via a mixing deck and amplifiers and into a PC that was connected via ISDN to the videoconferencing platform.
I had zero knowledge of any of this kit but it went in fairly simply and test calls were made - all successfully. I tried to find someone to sign it off but no one was around (or wanted to, from what I recall), so a call to my manager who said to have one more go then leave.
A few days later, same manager was on the phone shouting at me for a) leaving it in a non-working state and b) where was the sign off?
Hey ho. Back I went.
When I arrived, we tested it and it worked.
<sigh> here we go.
Then suddenly one of the directors came in to ask if I couldn't sort out her connection while I was stood around doing nothing else ( ! ). So, since the video conferencing stuff was working I decided I'd have a look.
Turned out she also used an ISDN line to get external connectivity and apparently it wouldn't initiate a connection, but when we walked into her office to check, of course it connected straight away.
Ok, well come get me if it stops working.
Back in the videoconferencing room, it suddenly couldn't get connectivity.
Hang on... yep, her office and the VC room shared the one ISDN connection so when one had initiated the connection the other was left without one.
Never got an apology but I *did* get a sign-off before I left.
My place of employment in the early 2000s was a small IT contractor in Toronto -- and by small, I mean fewer than a dozen employees. Our internet uplink was an ISDN line (BRI, i.e. 2B+D). This was presented as a little box screwed to the wall of the telecom closet, which had, as I recall, two pairs of screw terminals, one (or both?) of which was wired into whatever bit of our networking hardware, and an RJ-11 jack which went unused but was part of the telco's standard ISDN termination unit.
Well, twenty years ago tomorrow -- Thu, Aug. 14, 2003 -- at a little after 4 PM, the power went out in our office -- and, as we didn't yet realize, across much of north-eastern North America.
Once the outage had proved itself not to be transient, people started wanting to contact their spouses etc. Problem: no power = no phones; the PBX was out. (Someone will surely comment about battery backup. I don't remember the details, just the unhappy end result.) Cell phones also weren't happening. (I don't recall those details either. Maybe the cell towers were out too, but more likely they were simply massively overloaded as everyone in range tried to place calls.)
Fortunately, I remembered having seen a POTS phone sitting in a box in a closet -- not one of the fancy units that worked with the dead PBX, but just a standard, residential-grade single-line telephone. I plugged it into the empty RJ-11 jack on our internet uplink and woohoo! Dial tone! So people could take turns going into the wiring closet / makeshift phone booth, to reach whoever they had to reach. Or try to, depending on circumstances at the other end.
(The rest of this has nothing to do with ISDN; it's just about the blackout in general.)
By the time all the calls had been made, I guess we knew it was a major blackout and that no more work would be done that day, for we all decamped to a nearby restaurant. The only food they could serve was prepackaged stuff like potato chips -- no way to cook -- but this basement pub's claim to fame was their extensive beer menu, and that was all we were after. We spent an hour or so drinking by candlelight, then headed off in our various directions.
I had taken the subway (RightPondian: tube) to work, but both that and the (electrically powered) streetcar system were down, and the replacement buses threatened to be madness. So I set off on the looong walk home. (Google Maps tells me it was 10 km.)
I passed through the Bay Street business district (think Wall Street or the City). The traffic lights were all dead, but there were stockbroker/lawyer types at all the major intersections directing traffic. The above-linked Wikipedia article says the police issued some of them with high-vis vests, but I don't recall that; it must have happened after I passed through. I just remember Bay-Street guys in shirts and ties, playing amateur traffic cop.
Later, by which time night had fallen, my route took me along College Street through Little Italy, an area of restaurants and bars, many of which have patios. They were all candle-lit and all packed. None of them serving food either, I presume, but doing a roaring business in drinks.
Then, bonus! A pizza place was actually in full operation. Gas-fired pizza oven, I guess. I bought a slice or two, and that was dinner.
It seems most places got power back later that night, but in (my part of) Toronto, it was out for a couple of days. I don't remember much about the rest of the blackout. The one thing I do remember is the novelty of seeing So Many Stars at night in my urban neighbourhood.
Side note: my parents never lost power. They lived in a suburban neighbourhood in a small city an hour and a half from Toronto. They were in the middle of the blackout area, but near a couple of small local power plants, which I guess managed to isolate themselves the way every power plant should have done, and so avoided the general cascade failure. My folks only learned that there was a blackout when my uncle called from California to ask how they were faring.
Events such as these are why I decry phone companies' policy of eliminating landline, POTS payphones. It is a necessary function for citizens to be able to reliably summon police, fire, and emergency medical services, even when, and especially when, the shit hits the fan.
As demonstrated, cellphones and VOIP phones don't get that job done.
Agreed, less than a decade ago I worked on the telecoms provision for a new-build hospital, the main connectivity was (new to the organisation at the time) SIP but I still had to install a surprising number of POTS lines. These were needed for emergency links to blood transfusion service, ambulance centre, lift alarm panel etc. I left that field now but whenever I hear about the death of POTS I wonder what they are going to replace these critical things with...
I love the fact that POTS had it's own redundant power supply. I've maintained (probably at great cost to myself, although VM tell me I'll pay more if I remove it) a POTS landline for the entire time I've had this house - although I know it's soon to be going away.
It's like analogue radio and the push to digitise - yes you can cram more in to the same space, but the KISS principle should remain for anything that might be needed as an emergency comms medium.....
We had a complete technophobe of a marketing director, back in around 2002, who never used his computer. He'd always come to us to dictate emails (despite none of us being secretaries), and when we asked him why, he complained that his computer was really old and slow. And to be fair, it was the last 486 in the building, so we ordered him a shiny new Pentium 3.
When it arrived, I took out his old computer, and he took the opportunity to rearrange his office a bit. Meaning that the new computer was now too far from the network port, and I needed to get a new cable. We didn't have any of the right length in stock, so I ordered some new ones, and told him that I'd be back to install the cable in the next couple of days.
Things happened, there was a weekend, and it was 6 weeks before he came back to me to say that his new computer didn't work... SIX WEEKS!
...a thinnet BNC network terminator looks almost exactly like a BNC -> RCA adapter - especially if you have one of those little protective dust covers over the end of it. Back in the early 90s when I was in 6th form us poor comp-sci A level students (there were about 5 of us total, we had a rota) were often left in charge of the computing suite at lunch time so the actual staff could get some much needed rest, and the younger kids could use the computers to play games on.
If ever things got a bit much, or if we were particularly hung over (it was the 90s, no one would ID you as long as you didn't cause any trouble) we'd replace the network terminator for a nice quiet afternoon. "Sorry lads, network's down. Nothing we can do about it until the teachers get back and let us into the server room." Without access to the shared drive full of games most of the kids would just leave.
Many, many times in the past have I wedged a Euro-type Shuko or two-pin plug into a UK socket, even on occasion using a small screwdriver in the Earth-pin hole to open the socket's internal safety shutters.
Take that, MK electrical safety engineers!
OTOH, I have here in my office now several of those extremely sketchy Chinese-made "will accept any plug from any country" power-strips, and those give me the heebie-jeebies. Arcy-sparky and wibbly-wobbly don't mix with mains voltages.
A long time ago, when round pin sockets were still lingering in older UK houses I had a unversal plug. There was a slider adjacent to the flex entry point which would uncover different sets of holes in the plug face to let different sets of pins fall through - I think the selected pins could be rotated to lock them into place.
It was very clever except for one detail. It didn't have a conventional cable clamp, maybe it would have got in the way of the slider. It had plastic fitting which just clamped round the cable with a self tapper so that pulling the cable would have jammed the fitting up against the side of the plug and held it firm that way. The self tapper was so positioned that it could easily bridge the line and neutral....
wedged a Euro-type Shuko or two-pin plug into a UK socket
Please don't do it again.
Not only could you damage the shutter mechanism by forcing it, but wrong size and shape pins into wrong size and shape holes is asking for long-term trouble, even if it works in the short-term. You risk bending the electrical contacts and making things less safe for the next user, even if they use the correct plug. Loose connections risk poor contact and possibly arcing, especially with heavy "wall warts" or with high-power appliances.
I have also seen sockets abused in this way with safety shutters broken, either leaving the socket unusable (jammed shut) or unprotected (jammed open).
On a point of detail, MK sockets often have shutters that do not use the "normal" method of lever-in-the-earth-socket. In the past it was a kind of rotating thing that didn't require an earth pin at all, more recently they have had split or narrow levers which are much more difficult to open with a screwdriver, and combine the two methods to make it as difficult as possible to open the socket without three standard pins of standard sizes, shapes and spacings.
And on a point of pedantry, it isn't "MK engineers" who designed this system - it is a British Standard. See also Fatally Flawed and PlugSafe.
M.
There's a particular type of European or maybe American plug that is just the right spacing to fit the live and neutral of a UK 13 amp socket.
For many years, I had my Realistic (a Tandy brand) HiFi receiver plugged directly into a power bar (I would not do this using a wall socket) without a two pin adapter. I don't know why I didn't put a UK plug on the end, but I didn't.
I reasoned it thus. The receiver was not earthed any way, so even using a two pin adapter would not have used the earth, and would have made the connection looser, the plug had insulated shafts, so even if it pulled out a little (and this did not happen because there was an indent on the pins that made this difficult with the power bar I was using), it would not have exposed live contacts, the plug covered the remainder of the rectangular holes so it was not easy to push anything else into the contacts, and the power bar was in the back of the HiFi rack out of sight and away from prying fingers.
The only thing that could have happened was if something went into the earth, but that should not be dangerous anyway, and the power bar was fused.
I know that it was wrong, but it was too convenient. As a side note, as the receiver was not originally designed for the UK market, it also had an unswitched US style unshuttered and unearthed power socket for plugging something like a turntable in, which was covered by a plastic cover held in place with a sticker saying "Not for use in the UK". I was always a bit dubious about the legality of this, as it would have been easy enough to remover the cover and expose the unshuttered, unswitched live and neutral recessed contacts. I wonder ho many people actually did!
Used to do some tech'ing at a Theatre, and in the FoH cab which help the EQs, effects and assorted sound processing equipment, something had a two pin euro plug which someone had inserted into a 4-gang extension with the aid of a screwdriver, then withdrawn the screwdriver so the safety shutter now held the plug in place. For some reason, this was never given a fail sticker by the PAT monkey who failed pretty much every other electrical item in the building! (including failing class 2 earthless equipment for not having earth connectors... Sometimes, the phrase "competent person" isn't really that accurate)
Indeed - BBC Micro: "video doesn't work, I've plugged in the aerial lead solidly, it should work".
How the heck they got the monitor lead (DIN, RGB) rammed hard enough onto the BNC (Video Out) socket, so that it was, indeed, solidly in place, I do not want to know.
Luckily, DIN leads can be replaced and the BNC on the back of the Beeb didn't seem to have suffered at all.
Is there any other way?
My home printer is in a cabinet, so to connect the power and network you have put a loop in the cables near the plugs, push those loops through the hole in the back of the cabinet, put the cabinet back against the wall and then plug the cables in. now, pick up the printer in one hand, slide it into the cabinet far enough to get both cables connected, reach under the printer with your other hand and find said cables and connect them to the printer, remove your hand and push the printer all of the way into the cabinet.
it works...for a single user who knows how painful it is to remove and reinstall...but the other family members think that running used paper through it is de rigeur... I can't complain, because if I do I'll have to buy a printer stand.
That's like saying a USB type A connector isn't compatible with an RJ45 Ethernet socket.
Or the occasion when an exhibitor at an event here absolutely insisted that they needed a wired internet connection and couldn't rely on our (admittedly, pretty poor at the time) WiFi.
So I duly ran a cable and provided a small desktop switch and two RJ45 leads, only to be called 15 minutes later by a very cross exhibitor who insisted I'd provided "the wrong type of plug". While their Windows laptop was connected and working just fine, their Mac did not have an RJ45 socket at all, and apparently "this computer's network socket is that one there"... pointing at a standard USB-A.
M.
N connectors and BNCs. Well they may not be fully compatible but if you are in a pinch, you CAN connect a male N connector into a female BNC socket. It is a bit unstable, but it can work in a pinch. Of course don't try this when the N connector is an over 100 watt transmitter, and the BNC socket is a VERY sensitive receiver. I suspect that it might work in a networking sense, but it is fragile.
Probably too many stories, and I've already told some of them! They were the bane of our life in the mid-late 80's and the 90's.
Working at the local college with RMNET the little s*ds soon realised that loosening the BNC connector but not actually unplugging it gave an unreliable network for the next class and we (or at least the colleague who dealt with that side, I was largely a VAX man) would spend ages checking every single connector in the room.
I won't even go onto SCSI terminators on external disk drives - too painful!
In a later role as an IS Manager my DP Manager had a thing about never removing a terminator from a 5250 terminal. If the chain ran from A - B - C and he wanted to add terminal D he'd always disconnect B to C and add in extra cables using male-male connectors so the chain was A - B - D - C, even if physically D was located beyond C. An hour's rewiring in the main admin office removed about 20 cables and several cable bridges where he'd cabled across gangways totally unnecessarily.
The good news is he put the terminator back. Where I worked terminators would all go to the same place as biro tops. The spares would be half-inched by staff 'just-in-case'.
"A terminator, a terminator - my network for a terminator" was the eternal cry of the DRS20 Microlan engineer.
"Re-reading that jolted my memory of a desperate engineer actually soldering a bare resistor across the end of a T-piece. "
I have done that when stuck in the back of beyond and needed to get a network up.
I've also made terminators, a 1/4W 47 or 51R resistor fits very nicely into the centre of a BNC crimp connector, crimp on the sleeve and whack a bit of self adhesive heatshrink over it, you're golden.
"With the best little whorehouse too..."
Nope, that's in La Grange. Besides the movie, that brothel has a song about it, "La Grange" by ZZ Top.
As it happens, Paris, Texas also has a movie named after it[*], with a wonderfully evocative soundtrack by Ry Cooder.
[*] which I haven't seen, so no review is implied, but FWIW, it won a bunch of awards
"They were the bane of our life in the mid-late 80's and the 90's."
Oh yeah!
In our case it wasn't the terminators that tended to cause us pain, but the cables' BNC connectors -- not malice, but sheer flakiness. We had what I believe was the right crimper, but I never figured out how to install a BNC robustly. And in those (effectively) pre-internet days, there was no obvious way to learn. Or maybe the parts we had were simply crap.
"The network's down" was an all-too-common cry throughout the office, the cue for me or my colleague to spring into action. The obvious approach was bisection -- break the network at a T and drop in a terminator; see if that brings back the upstream section of the network; iterate. But that risked fubarring that BNC connector too, and so multiplying the failures.
I was a bit dubious about the newfangled 10base-T stuff (star topology seemed wasteful), but quickly learned what a huge improvement it was over thin-wire Heisen-net.
Worked at a university college in the early 2000s where A Certain Academic had the habit of using the terminator that exited by his desk at a convenient height as a coat hook. Over time, the terminator would loosen from the movement of his jacket being hung and removed.
Cue about once a year or so of screams from an entire building as it went offline when Prof X donned his jacket and went to lunch, or on one fun occasion, left late one day, and no-one noticed until the next morning. He did eventually stroll in the next morning before one of the facilities custodians turned up with the office key, but only by minutes.
Company where I used to work we had different networks for staff (green) and visitors (blue) - the network ports were colour coded to aid selection. One day all the networking for staff and guests went haywire. Took a while to find out that someone, whilst bored in a meeting, had used a patch cable to link the two networks together...
We once had someone decide that they wanted to connect their phone to a network and plugged in a D-Link router. This muppet didn't realise they plugged in the wrong port and the router's DHCP tried spraying out IP addresses; QIP then crapped itself and gave up meaning the whole subnet went down...
Icon 'cos scientists are supposed to know better...
"BTW, thicknet cable with N connectors makes dandy ham radio feedline. It's almost exactly RG-214, but with a solid center conductor. Fine for HF, OK for VHF and not so great for UHF."
Yes it does and I've got a promise of a couple of hundred metres of it at the next big cable tidy up.
A PA person for management was in a meeting room, they looped a Cisco IP 7970 phone back into an active wall port (They have a pass through port so you can share a connection, the Polycom based Cisco 7937 conference phone had broken so it was in as a spare). The network was configured without loopback protection/spanning tree enabled (Although the first time I occured I Googled and forwarded the fact the switches could prevent this sort of thing, but what did I know).
Absolute Chaos as a broadcast storm immediatly occured. Queue not being able to telnet on to the core edges switches as they were now overloaded so had to resort to the "Have you plugged anything in today" routine, realised there was one meeting room I couldn't get into.
Of course it was that room, unplugged the patch and all back to normal until the next day when there was a shrek of we can't work as no network.
"the Polycom based Cisco 7937 conference phone"
I never encountered the Cisco version, but Polycom's analogue version was (is?) a great phone -- far better audio quality than just putting a regular phone on speaker (at the local end anyway; I can't vouch for the remote end). Its styling is way cooler, too (the above links are to images).
Polycom's video-conferencing codecs, on the other hand, not so much (at least, as of 2009). No idea how good or bad they were at their core function, but if you wanted to automate a fleet of them, they were a major pain -- endless small variations in their command sets that the code had to account for. Unlike the Tandberg units, for which you could write write the code once and it would work on all of them.
Once worked at a company where they had Ethernet and Token Ring networks, and ALL the network ports were RJ-45. Not labelled, either.
Mistakenly plug an Ethernet PC into a Token Ring port in an isolated office and listen to the screaming from the System/36, AS/400 and Amdahl goombahs. Unplug it and slink away, never telling anyone.
Good times!!
It's really weird watching old episodes of computer chronicles, and they are saying stuff like "Will the future be XEROX's ethernet or IBM's token ring? Nobody can say for sure".
And you sit there wondering what possible technical reason could lead people to actually choose token ring. I know that hindsight is 20/20, but I don't think it took a genius to see how much better ethernet was.
As I recall from those dim and distant days the argument was that TR was deterministic, i.e. you "knew" that the data packet would get to its destination whereas ethernet was merely statistically probable. In real life this just meant unnecessary overhead on TR but the banking, legal and health types seemed to think it was the killer argument.
You also knew when the packet would arrive. For some applications, e.g. real-time process control, that matters.
How much better than token ring was the Ethernet of the day, i.e. 80s-vintage coax? (I've worked with thin-wire Ethernet, but not with thick-wire or with token ring, so this question is serious, not rhetorical.)
Sure, UTP's star topology is more robust than a daisy-chain of whatever flavour, but back when TR and Ethernet were duking it out -- and, I presume, when that Computer Chronicles episode aired -- 10BASE-T was still in the future, wasn't it? So can't have figured into their wondering.
Back in the 80s, the choice between TR and Ethernet was between two similar topologies (one a ring, one a bus, but both daisy chains, with all the problems those entail), and each had technical strengths and weaknesses. As I understand it, in theory there were legitimate technical reasons to choose either one, depending on your requirements -- Ethernet for better network utilization, but TR for guaranteed transmission times at the cost of fewer packets per second overall).
But that's theory. In practice, maybe TR really was so awful, even compared to 1980s Ethernet, that there should have been no comparison. What's the experience of people who have worked with both?
None of this is to discount the marketing BS. Sure, IBM would try to sell you their solution even if it wasn't the best fit for your use case -- and, being IBM, might well succeed. But that doesn't mean it was never the best fit, right?
BTW, if you're streaming one of those Computer Chronicles episodes and it stalls, well, you've just discovered why people building systems to control heavy machinery might care more about guaranteed arrival times than bandwidth...
Maybe it was because I was only using IBM TR kit, but in general I found it reasonably problem free, as long as the cable quality was up to spec. and you had control of everything that was plugged in (see elsewhere in these comments).
But the problem with 4Mb/s and the original 16Mb/s, using MAUs was that a MAU was dumb, the only smarts was the relay that looped in the cable plugged in to the port when power was applied.
But this meant that it would loop in a system running at the wrong speed, and stop the ring.
When CAUs came in, TR became much more resilient to user error.
My thoughts are that TR was physically more versatile than 10base5, you could extend the ring by plugging in more MAUs with less disruption than extending 10base5, and you could use a structured cabling system, as long as you kept to spec. A 10base5/10base2 mix was better than 10base5 with drop cables and AUIs, but you had to be a little careful about just how you branched the network. 10baseT changed everything.
TR was supposed to degrade more gracefully under load. Above a certain load, bused Ethernet tended to completely collapse as CSMA/CD kept getting collisions, regardless of how long it backed off. Again, with 10baseT and switches (rather than hubs), Ethernet behaved better, end eventually won the 'war'.
I did work on TR over (shielded?) twisted pair and Madge CAUs as a mere user, but never really had any problems that I noticed.
My favourite was the even-more-unwieldy-than-Ethernet-AUI-cables and the clumsy connectors with the "guaranteed to fail" bypass switches built in. I often wonder if IBM wanted to see just how much inconvenience their TR users would put up with.
I will grant that Token Ring, with all IBM components, did actually work. But, as soon as it got uppity, trying to use UTP like 10BASE-T, Bad Things began to happen. The low jitter requirement (and its consequent sharp voltage transitions) wreaked havoc at EMC testing time, and while 4 megabit TR could, with some effort, be called compliant, getting 16 meg UTP TR switches to run reliably AND pass EMC testing often called for a case of beer to be delivered to the EMC test technician, and his attention directed elsewhere as the system scanned through certain frequencies.
When 100BASE-T arrived, TR died a quick death, and was not mourned. The cost for a 4/16 TR interface card, IIRC was about 3x that of an Ethernet UTP card. Some of that was due to licensing required by TR and not by Ethernet, but the TR MAC also relied on downloaded firmware, while the Ethernet MAC had a much simpler state machine built into the chips.
I developed a low cost TR 4/16 interface card for Data General's AViiON workstations, and some TR switches for 3Com, but I prefer to pretend the entire experience never happened.
And you sit there wondering what possible technical reason could lead people to actually choose token ring
Perhaps the simple fact that it was faster than 10M ethernet (which was a common speed of the day) AND it would hold that speed when heavily loaded.
TR was 16Mbps IIRC, and because it was deterministic it would do 16Mbps regardless of how hard you tried to load it - all that happened if you tried to do too much was that everything had to queue a bit.
Back in those days, switched ethernet was a rarity - 10 and 100baseT were effectively just a twisted pair conversion of 10base2 (or 10base5) with all the same contention issues. If you search, you'll find that these max out (depending on the traffic mix) at about 40% of theoretical speed due to collisions, and if you try to push past that then throughput actually goes down. So with 10baseT your max throughput is around 4Mbps, with 100baseT it's around 40Mbps and if you push it then it goes down - with TR it's 16Mbps all the way regardless.
Also, planning was simpler with TR. In effect (IIRC), as long as you didn't create a cable loop then you just plug everything together and it'll work - in practical terms no limit on ring length, just on branch length as many of the hubs were active repeaters. With ethernet you had to follow the 5-4-3 rule - max 5 segments in total between any two nodes, max 4 repeaters, max of 3 segments populated (meaning that in a 10base2 or 10base5 network, some of your inter-repeater segments had to be point-point links and could not have any other nodes on them). And each segment had a max length (185m for 10base2, 500m for 10base5, 100m including patch leads for 10/100baseT) which could make large networks "interesting" to design unless you had an endless budget to allow fibre optics.
TR was, if you used the correct connectors, fairly idiot proof as well. Pull a plug out, and either the wall socket provides a loopback, or the hub at the other end detects the disconnected device and automatically bypasses that port.
There were two things that killed TR. Firstly, being a predominantly IBM system it was "reassuringly expensive" - basically nothing was cheap with TR. That alone was probably enough to kill it outside of certain markets. But also, the cable was quite bulky - less so that the '1/2" hosepipe' of 10base5 cable, but more so than the thinner 10base2, and even more so than the UTP starting to roll out for ethernet. Of course, with 10base2 or 5, you only needed one cable for multiple devices - but you had the potential for a simple fault to take out the entire network as described. And the cable was itself "reassuringly expensive".
I strongly suspect that had IBM (I assume it was IBM with the patents and licensing costs) been more astute, then TR could have won out. Until switches started replacing hubs in ethernet networks, there was really no technical reason for choosing ethernet (particularly 10/100baseT) other than cost.
And roll forward a few decades (now that make me feel old), and you can throw an ethernet network together with a pile of switches and short of having loops "it'll just work". No problems with the size of collision domains (which was the reason for the 5-4-3 rules). Kids today just don't know they're born ...
We had RJ-45 jacks for "commander's console" style AT&T phones (usually on departmental head secretaries' desks), for ethernet, for token ring (with central "star" switches) and for hydra connections. What're those, you ask? Hydra connections were a way of using ethernet cables to carry ethernet, which technically only requires four wires, and, simultaneously, AppleTalk on the four "unused" wires. It worked acceptably well, back when ethernet was running at 10Mb/S. "Hydra head" cables had an RJ-45 on the wall end, and two connectors on the (laser) printer end. The AppleTalk connector went to the AppleTalk interface in the printer, and the ethernet connector went into the ethernet printer interface.
Somehow, some way, the users were trained (or beaten into?) usually not plugging things in willy-nilly, and instead, calling Telecom/IT to do it.
"You could also use RJ45 for RS232C serial communications"
Or even RJ-11s, if all you needed was TxD, RxD, and ground (i.e. software flow control). Where I worked in the 90s, the serial-terminal building wiring was done that way. Each terminal (Wyse'es, model number long forgotten) had one of these little adapters plugged into its DB25. An off-the-shelf phone cord went from that to the wall-mounted RJ-11.
Unfortunately, we had two offices, whose in-wall cabling patterns were different (one was straight-through, I think; can't recall what the other was) -- which meant that the adapters we wired up for office A wouldn't work in office B and vice versa. There came a point that I got so frustrated that I remade a bunch of connectors at the server end to compensate, so that all the user-end wall-mount RJ-11s ended up with the same pinout.
Then we expanded again, and the new space had to be renovated, including data cabling for yet more serial terminals. The price difference between phone quad, which is what was called for, and CAT-5 was small enough that I convinced management to spring for the extra. Drastic overkill for the purpose, but "we'll need networking one day". (Can't recall for sure, but I presume we made up cables with an RJ-45 at one end and an RJ-11 at the other, wired so that the RJ-11 presented our standard pinout.)
Sure enough, within a year or two, the Wyse'es were gone, replaced by networked Win95 boxen. One of those all-too-rare instances where YAGNI (you aren't gonna need it) proved false, and foresight paid off.
(Now that expansion space is a pet store, and all our old cabling has presumably been ripped out.)
Token ring had a massive technical advantage over ethernet before ethernet switches became a thing: there were no collisions. You load up a 4Mb token ring and it would sit there all day with 4Mb/sec throughput. Attach lots of chatty clients to a non-switched ethernet and it would collapse under all the collisions. Sadly, it was a bit like the beta vs VHS thing. IBM’s proprietary token ring cost big $, because… IBM. Ethernet was cheap and “good enough” for most applications.
I once had an interesting technical issue converting a building from the late 90s token ring over UTP to 100Mb switched ethernet.
Take a wall outlet that was working perfectly on token ring, patch it into the ethernet switch and the shiny new ethernet card on a PC… didn’t work. Swap back to token ring on the same Cat 5 cable… worked perfectly. Cue much scratching of the head when the same thing happened in the next office over (and the next…).
Bearing in mind that this happened in Australia, where any electrical work is over-regulated beyond belief and even back then, you needed a special license even to run a pre-terminated patch lead through a wall cavity… Every wall outlet was wired using the 568A wire colour codes. Every outlet at the patch panel had been wired as 568B. The result: every wall outlet was a crossover cable. A token ring network didn’t care, but ethernet back then did care.
I have an anecdote about Token Ring. I worked at a startup in the 90s, and we decided to develop a Token Ring hub. (The company was reasonably successful making other networking gear). During development, we moved the company to a new building. The decision was made to deploy 100Mbps ethernet for all computers in the new building, even though that is what we were competing against in the market. LOL. In retrospect, this was the right call. This was back in the days when Synoptics was a leader in ethernet.
I thought it was ironic at the time that this decision was made. The main reason was price. But what about deterministic performance? "Yeah, we're not paying 2x for something you can't explain to me". Enet was cheaper, and TR was prone to issues at that time (jitter aggregation). The product was ultimately a failure.
I used to work in network operations for a large IT outsourcer when that new fangled NT 4.0 came out. One morning we started getting calls saying no one could access the Notes servers or the internet and rudimentary checking over the phone showed they had IP addresses but none that matched our range.
A stroll around the building found a bunch of young server admins on a training course for the new wonder OS with a server on the floor brought in by the instructor. A pair of wire cutters through his patch lead soon removed his errant DHCP server and after a reboot our users returned to their paying work. Strong words about competence to instruct were had.
We had a single private 16Mb/s token ring running around a major support centre. It was managed by the global comms support group of said company.
Now, in case you are a bit unfamiliar with token ring hardware and IBM's networking infrastructure, it appeared at the wall as a large, square socket, which could actually host several different protocols, including 3270 terminal access.
In a week of major disruption, said token ring would break (with beacon packets being broadcast), but although we could logically see which station was either side of the break, because we did not manage the infrastructure, we could not work out what was happening.
We appealed to the comms team to help us, but they could not see the problem, as their documentation showed only our desks wired to the ring.
Being told the desk wiring was good, we ended up suspecting a faulty TR card, and literally ended up out-of-hours turning everything off, then turning it on one station at a time. But everything came back fine!
Until a few days later when it happened again, and again a few days later.
We could not find a culprit.
Eventually, a slightly more open comms person decided to check their documentation against what was actually in the network rooms and risers. And lo and behold found a rogue cable plugged into one of our MAUs in the main comms room. The cable was not labeled, but was finally chased down to one of the conference rooms, where it was found that some regular visitor had been plugging their 4Mb/s computer into our 16Mb/s ring, and not getting it to work (but breaking the ring!) but leaving it plugged in until they left for the day.
Well, our management could not face the same thing happening again, so we went through an operation of breaking the ring in two, and putting a bridge between the two segments, with the main servers on both segments, so that if it happened again, only half of the call centre would be affected.
I was shown the comms room where the problem had occured at a later time when we were vacating the building. The wiring was beyond a twisted mess of cables, and I was told that the situation was so bad that instead of re-using cables, they were left in place and new ones used whenever they had to make changes. The only wonder is that there were not more similar incidents.
This post has been deleted by its author
Can't make my mind up if you've missed the joke icon or don't know what the terminator does.
In case it's the latter, whenever you send a signal down a cable - at any change in impedance there will be a reflection of some sort. An open ended cable is such a change in impedance, and will result in the signal bouncing off the end of the cable and flowing back up the cable. Thus every transmission ends up interfering with itself after a delay (twice the propagation time from the signal source to the end of the cable).
The terminator (in this case) is a simple resistor that matches the impedance of the cable and effectively just "swallows" the signal that would otherwise reflect.
In the '00s we had an outside visitor who would just pop out an existing network lead from a wall socket and plug his laptop in.
# How do I get a connection?
$ Enter your user name and password in the connection dialog.
# What's my user name and password?
$ Well, your user name is your initials followed by your payroll number. *You* know *your* password.
# What's my payroll number?
$ It's on your payslip
# But I don't work here.
$ So why are you trying to access our network?
# Can I just log on as you?
$ Well, a) No, because I would then be fired, and b) No, because your laptop's network address won't be registered as one of our computers and won't be allowed access.
We once had a colleague, let him be known as Anon, who knew much more than we did about electricity, so when he found that none of the Thinnet cables that arrived in his lab had properly earthed shields, he carefully soldered them all to earth (or ground, as it's known in Umrika). And then he told us that our Ethernet had an unacceptably high error rate despite his improvements.
We then explained the concept of earth loops to Anon and removed all his helpful solder. With all the 50 Hz electricity gone, things worked much better.
Sort of similar story from somewhere I used to work.
I got a call from one of the managers that her computer had stopped working - more specifically, the terminal session to the main business system had "just stopped" responding. I took a wander down to her office, to find a visitor sat at her meeting desk - with a network cable plugged into the laptop.
The manager didn't need much comment on the lack of network cable in the back of her desktop to realise what the problem was.
About 30 years ago I expanded a thin=net network installation to a total of about 10 computers. A few months later, I got a call that their network was suddenly very unreliable. When I got to the location, things were not as they were when I completed the original expansion. When they moved their home-made cubicles, they pulled the thin-net cable through the small openings in the cubicles. Some of the cables caught on the wood or other obstructions and strippe feet of insulation from the cable and damaging much of the braided shielding as well. I was surprised the network worked at all. I replaced the bad sections of cable and all was well again.
I worked for a well known computer company as on-site tech support in a VERY well known airline main office. This office did all the ticket requests for the whole world (I think). The net was spread over 4 floors and was thicknet with delay boxes (If I ever get hold of the guy who designed it ....). Problem was if the connection went down, (TThere was 6 separate lines), and you had to start on the end of the dead loop. Each line went in and out over the floors in no particular order.
I lost weight (Thanks, I needed the help) in my first three months going floor to floor running up stairs, down corridors, completely around the floor to othet departments, etc.
After 3 months I had the new working, knew the layout better, and had everything in hand. Until they decided to mod tghe server beast. That's when I got out. I was to be on 24/7 call with no extra pay.
Miss those days. The big city I worked in had a lot of small stores in the area, and plenty of parts places.
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
