- Feature Articles
- CodeSOD
- Error'd
-
Forums
-
Other Articles
- Random Article
- Alex's Soapbox
- Announcements
- Best of…
- Best of Email
- Best of the Sidebar
- Bring Your Own Code
- Coded Smorgasbord
- Mandatory Fun Day
- Off Topic
- Representative Line
- News Roundup
- Editor's Soapbox
- Software on the Rocks
- Souvenir Potpourri
- Sponsor Post
- Tales from the Interview
- The Daily WTF: Live
- Virtudyne
Admin
The real WTF is that Karen didn't have the pride in her job to do it right. I would have explained the situation to them, at the very least, and let them make the decision. If they didn't want to, then you ride our your contract peacefully and quietly. But rolling over on your principles without even a peep? Forget it.
Admin
So she converted it to Token Ring?
Admin
Oh, come on! No documentation? Put together by one guy who left (probably for twice the money)? Failing hourly?
This was state of the art before Al Gore gave us the Information Superhighway. You kids don't know how good you have it.
Admin
This, of course, lead me to think of how to recover such a situation. The only thing I can think of would be going back-ish to thinnet, using two UTP-to-coax baulins at each system. At that time, most, if not all, adapters had thinnet connectors as well. How horrid. Well, maybe some places would allow you to put multiple systems on a hub, and a few switches making new collision domains. Six months with that thing around my neck. ~Shudder~
Admin
The decision wasn't hard to make at all...so what was it? Do I have to read the stupid HTML source to find out?
Admin
Of course, this also means that all the cables were wired as crossover cables, instead of the straight-through you'd want.
So in the end, you have a redundant NIC on each machine, crossover cables everywhere (at the least you'd have to replace one connector per cable), and a pile of switches to buy.
Nobody's going to approve that. If only you could sell the extra NICs to buy switches...
Admin
Admin
I've been wondering how you would fix this, as well. The only thing that comes to mind (keeping in mind that I'm not a networking engineer at ALL) would be to invest in a bunch of switches/hubs and try to segregate as much of the "network" as possible, so that if someone did reboot their machine they'd only take down that office/building/whatever.
Admin
Next step: Call the vendor who sold the network gear. Discuss problem with them for two weeks. Eventually they'll propose to bring out "The Lanalyzer" which is a $10,000 gizmo that we would replace today with a laptop running Wireshark. "The Lanalyzer" comes with an expert who is the only one who can operate it or interpret the results.
He looks puzzled for awhile, then recommends that you purchase a hub ($5,000) switch ($10,000) or bridge ($20,000). When you ask why, he says to fix your problem. Which is? That you need a hubswitchbridge. Well how does that work? It fixes network problems. By doing what? Ummm, well, it uh fixifies stuff. OK so what's the difference between a hub and a switch and a bridge? Well a bridge is better of course (because it costs more).
Or to put it another way, you can't find anybody who can explain how things work. Just spend money and hope.
Admin
OP here. I'm not sure why my submission underwent a gender change. Anyway, the government contract in question made it abundantly clear that they were NOT going to play for more than a 2-month contract. So I did the best I could, and yes I did leave an apology note...unsigned, of course.
Admin
Reminds me of one of the first networks I worked on. We used the coax version of 10Mbit ethernet, and it worked fine, despite that we used 75R cable (video company) rather than 50R. It only went wrong when the MD took the computer on a trolley away to test something in the lab, and disconnected the coax from the T-Piece, leaving the terminator on the PC. He had been told multiple times that he should disconnect the T-piece from the PC, leaving the terminator at the end of the cable, but he was the MD, so why should he listen?
Admin
Hoping against hope, the place may have been wired with wall sockets. What little sanity that existed there may have been spent wiring the wall sockets straight, and then you would have had one crossed and one straight cable for each system. Pair the crossed with another crossed, and you've 'two wronged a right'! Still doesn't fix most of the problems.
Admin
Here's the hilarious part: you're not the submitter. You may have submitted a very similar story, but it's not the one I was looking at when I wrote this.
Actually, that's not hilarious at all- that's just depressing. How common is this crap?
Admin
I'm the submitter and so's my wife.
Admin
Can't find anyone who understands networking? But why would you? Any computer guy with pride would learn it himself, right?
Let's see... no Internet, so forget google and wikipedia.
Maybe you can get a book at the local bookstore... yeah, lots of stuff on DOS and CP/M... networking??? Try the fishing aisle. No, not phishing, we don't have that yet.
So you call the vendor. The one who charges $10,000 minimum for anything. You ask if they have a book. No, but they'll sell you a CD for $45 that includes an intro to networking.
So you buy it. Never mind that your computer's CD drive rarely works because there's some kind of BIOS interrupt conflict.
Oh, but you can't read the document on the CD. Hafta buy some reader software now.
Finally you get it open (5 minutes to load a page) and you discover it is basically a catalog of the vendor's parts. Still not how the hell it fucking works if indeed it ever does!
Ahhh, the good old days...
Admin
Because when they found out you'd failed to alert them to the problem, management had your balls?
Admin
Admin
Admin
Thank god for LaTeX!
Admin
Back in the early '90s, the small development group I was in (contractors for an Air Force human factors project) upgraded from sneaker-net to Ethernet. So we got some NICs and coax and went to the local computer parts shop for a couple of terminators. It. Didn't Work. We eventually got exactly two cards to work at the same time.
Turns out that the terminators were mislabeled. Using a multimeter showed that they had 75 ohm resistors in them. So we went back to the parts shop... where we found that the whole bin was 75 ohms. Apparently the little old lady who printed up labels and stuck them onto the packages had a senior moment. Those two NICs were just the only ones that could work with the mismatched impedance.
Admin
That's... inspired.
I would never have thought of coming up with that kind of "solution" to the problem. Hat's off to the person/people that came up with the original "design"!
Admin
No, you'll have to use your brain to figure it out. Provided that you have one.
Admin
Admin
I had the BNC connect NIC on my computer. Went to a LAN party and was basically told "That won't work." Technically, he was incorrect, I just put the connector on wrong (forgot the T-bridge and the terminator... which was attached to the T-bridge). But, alas, I was out $20 for a new NIC... or maybe it was the network cable (my NIC may have had both the BNC and the RJ-45 connector on it).
But, yeah, this was 1998.
Admin
This was 1995. Exciting times - cutting edge, green field, writing automated systems over office networks replacing filing cabinets.
Maybe what the previous developer made was WTF'y but getting things to work at all was the challenge.
I wonder what Karen is doing now...
Admin
Admin
Hi Bob,
I have to admit that as a female software engineer, I like reading tech stories with a female protagonist. It counteracts some of the misogynism you see sprinkled throughout the comments (and 3 ... 2 ... 1) ... but now, knowing the reality, I wonder if it's really a good idea to switch genders, even for a good cause. Maybe it just indicates that not enough women submit (good) stories, so Alex feels the need to spice things up a bit. Perhaps this is a call to action for all the female technical people out there - submit your stories! But I'll admit that I personally don't feel comfortable sharing my stories - I don't want them being connected to my company, which I feel overall does a really great job.
Admin
Admin
Admin
(GOD DAMMIT)
Admin
Again, just for clarity: while Bob may have submitted something similar to this story, it was not Bob's submission. The original protagonist did identify as female, although it was submitted anonymously, so the name is entirely invented.
We love submissions from anyone, but it'd be great to have a little more representation from the female readership. We actually try to avoid changing genders in stories, although sometimes we have to make our best guess because it's not clear from the submission.
Admin
I have never been a hardware expert and I don't know the technologies of setting up a network.
It does seem to me though that Karen shied away from being brave and explaining the exact costs of setting up and maintaining a network.
This would also have to be balanced against the advantages of having data automated and the costs saved from not being manually stored.
Obviously as we're talking about a large scale here, they might want to seek a further opinion.
Admin
And for the record, I used 10Base-2 on my home network for a very long time. A single cable is much easier to run between the basement office's 4 computers to the attic office's 4 computers, I didn't have to buy hubs, and 10 Mbit was fast enough.
--Joe
Admin
This comment was oddly very helpful, though it did nothing more than sum up the state of the technology field as outsiders interact with it. Very well written.
Admin
Admin
Admin
Around 1992 I worked in a place with 10base-2. My co-workers didn't understand about tees and terminators. They would do things like put a tee in a cable, then run a 20 foot cable from the tee to their computer and straight in to the NIC. This caused lots of hard-to-find errors, like most computers could talk to most others, but certain pairs wouldn't work.
Admin
I'm just amazed at how much more difficult it was for this to be done the wrong way than the right way. Getting Windows 3.x to forward packets between NICs, even if it was using NETBEUI, is a lot harder than plugging into hubs. The guy must have been upgrading from 10base2 and thought this newfangled 10baseT thing was just a ploy to sell twice as many NICs.
Admin
ObBeavisAndButthead: "Heh heh heh... he said, "Ramming it home"... heh heh."
Admin
I am a bit confused I thought token ring did daisy chain, so had this guy used two nics instead of a t-connector? Since the hardware was already installed would it have been possible to put in a second set of cables and have a dual token ring and a bit of redundancy for the price of extra cable and t connecters?
Admin
Admin
Obviously, Karen's predecessor upgraded from a 10-Base-2 network (using 50 ohm coaxial cable running from one node to the next and needing two 50 ohm terminators, one at each end) to a kind-of-10-Base-T-network (using twisted-pair telephone cable, supposedly star-wired from a central hub or switch), but in this particular configuration, he saved on hub/switches as well as cable length by running the cat-5-cable from node to node. This is actually quite creative.
There would certainly have been ways to ease the transition to a real 10-Base-T network by running a cable from a central switch to a decentralized local switch in every room, and hook up the computers within that room to the local switch. This to avoid having to run a cat-5-cable to the central switch for every single computer.
Admin
Cue "I'll be back" and "Resistance is futile" puns. Oh, the impudence!
Admin
Man, RISC architecture is going to change everything.
Admin
We had a similar system There were two networks - a secure one for classified stuff and a general one. But since everybody needed to have access to the secure network and to the general one with email.
The solution - two nics in each machine, turning every machine into a potential bridge to the secure network.
Admin
I'm the submitter!
Admin
Ethernet began life as a multidrop shared backbone using thick (10base5) or thin (10base2) coax cables. (There was a slower version before even that, 4baseSomething, but let's ignore that, OK?) 10base5 could be tapped live without breaking the cable. You drilled into the cable until you reached the centre, then attached an AUI transceiver to it. This had a pin that went in the hole to contact the centre conductor of the coax cable, and a sheath on the pin that contacted the signal shield conductor. You then connected the NIC via a cable to the transceiver. The ends of the cable were terminated with impedance-matching terminators.
10base2, thinwire, used 50-ohm coax, nasty skinny stiff stuff, with breaks for "T" pieces. Each T piece clipped directly on the NIC, and had either two cables or a cable and a terminator. The result is term, T-with-NIC, cable, T-with-NIC, cable, ..., T-with-NIC, term. Adding more machines was a source of stress, because it entailed inserting a new cable and NIC into the chain.
Both 10base2 and 10base5 had overall length limitations imposed by the CSMA/CD protocol used by ethernet.
10baseT is a twisted-pair adaptation of ethernet, normally involving straight-through cables from NIC to a central device (hub, switch, or occasionally a bridge - the difference between a bridge and a two-port switch isn't particularly obvious) in a star-shaped physical configuration. At the "logical" level, however, a hub-based 10baseT network still exactly resembles a 10base2/5 network. Switches and bridges complicate this somewhat by separating the physical network into multiple segments, but the basic idea is similar. Faster ethernet technologies are conceptually similar to 10baseT, so I'll ignore them for now.
Token Ring's conceptual design is a ring (not a daisy chain - there are no end stations in TR) of stations linked by cables. By the time I first experienced TR in action, it had moved the ring into a box in the network, with ugly ugly fat cables which had even uglier connectors on them linking the box to the stations. TR is a token-passing multiple-access network. Each station waits until it receives a "token" (magic packet) before sending. If it has nothing to send, it sends a copy of the token. All bits are allowed to circulate back to their sender. TR has deterministic behaviour in the face of heavy traffic, and is quite capable of reaching 100% utilisation for long periods, unlike non-switched Ethernet networks.
Later TR versions dispensed with the uglyfat cables in favour of normal unshielded twisted-pair cabling, but the ring remains. Multiple rings can be bridged together, complete with MAC-level source routing and hurricane-force broadcast storms from single packets in certain cases (the so-called All Routes Explorer frame, which will trivially reach a 500:1 multiplier with the right ring-to-ring topology).
The OP's network, however, had been set up using the PCs themselves as ethernet-to-ethernet bridges over 10baseT cabling instead of using a hub or switch. I just hope they didn't do something stupid like connect it in a ring, because I doubt that whatever wonky bridging layer they had in WfW3.11 could handle proper Spanning Tree protocols, and all broadcasts would therefore be received two times (clockwise and anticlockwise), thoroughly confusing the "which port is mac a:b:c:d:e:f on" part of the bridge.
Admin
Step 1: Open your browser's bookmark organiser. Step 2: Create a new bookmark with the following in the URL:
javascript:(function(){document.getElementsByTagName('body')[0].innerHTML=document.getElementsByTagName('body')[0].innerHTML.replace(/<!--/g,'<span%20style="background:#FF0000;%20color:#FFFFFF;">').replace(/-->/g,'');})();
Step 3: ???? Step 4: Profit!!! Uh, I mean click your bookmark rather than 'view source' when you want to read the comments.
Admin
In my younger days, I spent the better part of an entire day running down intermittent network problems in a thinwire network. The problem turned out to be a 50-ohm BNC terminator that went open-circuit for no good reason in the middle of the night. I kept it as a trophy, and am looking at it as I type.
The depressing thing is that the terminator had exactly one part (a 51 ohm resistor) inside. And that part went bad...
Admin