It still happens, the modern standards just handle these impairments better with line coding, active equalization, and more sensitive modem electronics. And obviously now we have shielding requirements in the PHY for 2.5G and above. But basically as DSP got cheaper and more compact, they were able to throw more compute at the problem which made a huge difference.
Most ethernet cable is UTP, literally "unshielded twisted pair." Shielded cable is much more expensive and less physically flexible due to the metal jackets, so people dont tend to buy it by default.
You can argue the jacket is shielding, but mostly ethernet cable is not shielded. The braiding will cause problems, but likely very minor ones based on the length of the the run that CRC will compensate for.
Yes, in theory. But in practice it depends on the frequencies.
Even if interference is insignificant, looping the wires around a metal core (which braiding does) creates a different impedance and can degrade the signal.
It's the wrapping of it around a ferrite core that does it. Surprisingly little can make a difference at certain clock rates. If you take a cat5 cable running 1G and loop it a dozen times in a 6" circle and put a metal screwdriver perpendicular to the loop in its center, you will see a spike in bit errors. Weaving one wire with another isn't going to be as strong of an effect, and it's fiendishly complicated in terms of an e/m problem, but it could easily be significant.
I hear you, but how often have you had to swap out a cable compared to how often have you had to deal with a rats nest? This isn't a whole lot different than zip tying groups of cables together.
IMHO this would be even better with 6 different colors per braid. Then you could trace a cable in a few seconds while also not having a rats nest.
Edit: oops, this was supposed to be a reply to another comment.
Velcro is almost always the way to go with cable management. Zip ties can easily damage the wire and encourage messy wiring since you have to cut and replace every one to change a single wire.
Wire lacing is OK for permanent installations, but good luck changing anything.
That's actually not that awful. Time consuming and you better not need to change the loom after it's done, but that doesn't become spaghetti that easily. At least not until someone adds few individual (and way too long) cables tangled to those.
I'd guess braiding is not that much more time consuming than fighting cables in a sleeve. Personally I'd use velcro-straps, but on our team there's always too little time to actually make things look decent, so pretty much all is in various stages of spaghetti, depending on which cabinet you're looking for.
unpopular opinion:
The effort of "tidying up" cables is never worth it.
You'll always end up having to change something and your sophisticated cable management turns a 5 minute job into a 1 hour job, as you open zip ties and dig out cables.
Just use the correct cable length, plug that shit in in a sensible order, label both ends with numbered tape, and let it look messy.
The first step is using correctly sized ethernet cables. Dont use a 12 ft cable for a 3 ft run. There shouldnt be more than a foot of slack in general. You dont need to "tidy" excess you dont have.
Then, for when you do need to tidy cables, use velcro ties, not zip ties. Ideally, you still label all the runs, especially the long ones. Seems a bit redundant for 3ft, but its still helpful.
At least put in the minimum effort of trying to follow the original path (assuming someone worth their salary actually was involved in the making of the site.).
Youd be surprised at the difference in mess from just plugging 'a to b' and 'a to b via c' no matter how many techs having been involved.
I'd obviously plan the initial setup, start with connecting ports close together via short cables, then put the longer connections on top, avoid unnecessary crossings.
And when switching out a cable, look for the most direct path
Or you could mark the bad one, replace the whole set, then unbraid and toss the marked one later in a dumb meeting you have better things to do than attend or something :p
43 Comments
mossberg590@lemmy.world · 58 pts · 163d
Sure, a long as those cables are shielded. Crosstalk was a real thing "back in my day".
SanctimoniousApe@lemmings.world · 27 pts · 163d
Exactly the first thought I had, fellow old-timer.
socsa@piefed.social · 17 pts · 163d
It still happens, the modern standards just handle these impairments better with line coding, active equalization, and more sensitive modem electronics. And obviously now we have shielding requirements in the PHY for 2.5G and above. But basically as DSP got cheaper and more compact, they were able to throw more compute at the problem which made a huge difference.
YiddishMcSquidish@lemmy.today · 7 pts · 163d
I don't think I've seen a single data cable that didn't have some kinda shielding since the early nineties.
rainwall@piefed.social · 21 pts · 163d
Most ethernet cable is UTP, literally "unshielded twisted pair." Shielded cable is much more expensive and less physically flexible due to the metal jackets, so people dont tend to buy it by default.
You can argue the jacket is shielding, but mostly ethernet cable is not shielded. The braiding will cause problems, but likely very minor ones based on the length of the the run that CRC will compensate for.
stupidcasey@lemmy.world · 3 pts · 163d
Would the twisted pairs prevent cross talk between two separate cables like it does individual wires?
CannonFodder@lemmy.world · 6 pts · 163d
Yes, in theory. But in practice it depends on the frequencies.
Even if interference is insignificant, looping the wires around a metal core (which braiding does) creates a different impedance and can degrade the signal.
YiddishMcSquidish@lemmy.today · 2 pts · 163d
Get out! From the inductance? How long it a length of wire does it take for that to become an issue?
CannonFodder@lemmy.world · 7 pts · 163d
It's the wrapping of it around a ferrite core that does it. Surprisingly little can make a difference at certain clock rates. If you take a cat5 cable running 1G and loop it a dozen times in a 6" circle and put a metal screwdriver perpendicular to the loop in its center, you will see a spike in bit errors. Weaving one wire with another isn't going to be as strong of an effect, and it's fiendishly complicated in terms of an e/m problem, but it could easily be significant.
YiddishMcSquidish@lemmy.today · 3 pts · 163d
Granted I'm no network engineer. But I have made many of my own cat5's through the years and I have not seen one since I was unable to legally drive.
socsa@piefed.social · 4 pts · 163d
The IEEE specs up to 2.5GBaseT do not specify any shielding for cabling.
YiddishMcSquidish@lemmy.today · 2 pts · 163d
That's fair. This is just my limited experience I'm speaking from, and I'm not the one buying the spools.
friend_of_satan@lemmy.world · 30 pts · 163d
I hear you, but how often have you had to swap out a cable compared to how often have you had to deal with a rats nest? This isn't a whole lot different than zip tying groups of cables together.
IMHO this would be even better with 6 different colors per braid. Then you could trace a cable in a few seconds while also not having a rats nest.
Edit: oops, this was supposed to be a reply to another comment.
alyqz@lemmy.sdf.org · 19 pts · 163d
Velcro is almost always the way to go with cable management. Zip ties can easily damage the wire and encourage messy wiring since you have to cut and replace every one to change a single wire.
Wire lacing is OK for permanent installations, but good luck changing anything.
Rooster326@programming.dev · 6 pts · 163d
Velcro costs so much more than zip ties.
IsoKiero@sopuli.xyz · 7 pts · 163d
Velcro is also reusable and removing it doesn't require sharp things around delicate and often pretty important cables.
DataCrime@lemmy.dbzer0.com · 2 pts · 163d
Also rough if you got cats 🐈 🪮
jaybone@lemmy.zip · 17 pts · 163d
But you can cut a zip tie. This is terrible if you have to replace a cable.
JAPHacake@feddit.uk · 2 pts · 163d
Or if you need to verify where each end is landed.
DataCrime@lemmy.dbzer0.com · 1 pts · 163d
Better than cable lacing 🤓
chaogomu@lemmy.world · 12 pts · 163d
You could also use tape tags on the ends, either color or number codes. Just mark both ends of each cable.
socsa@piefed.social · 21 pts · 163d
When you have automated most of your job, but you still take Adderall every day.
IsoKiero@sopuli.xyz · 19 pts · 163d
That's actually not that awful. Time consuming and you better not need to change the loom after it's done, but that doesn't become spaghetti that easily. At least not until someone adds few individual (and way too long) cables tangled to those.
socsa@piefed.social · 3 pts · 163d
It's awful considering braided cable sleeve is orders of magnitude cheaper than IT labor.
IsoKiero@sopuli.xyz · 2 pts · 163d
I'd guess braiding is not that much more time consuming than fighting cables in a sleeve. Personally I'd use velcro-straps, but on our team there's always too little time to actually make things look decent, so pretty much all is in various stages of spaghetti, depending on which cabinet you're looking for.
emotional_soup_88@programming.dev · 14 pts · 163d
If these are UTP, I demand a screenshot of
cat /proc/net/devand of some third party packet checking software that measures crosstalk and bit error. 😂anton@lemmy.blahaj.zone · 10 pts · 163d
I actually had to teach someone in my uni course how to braid power cables for a induction motor. Do they not teach basic skills in preschool anymore?
RamenJunkie@midwest.social · 9 pts · 163d
They only teach girls how to braid in pre school then ostracise them out of becoming power techs.
socsa@piefed.social · 3 pts · 163d
We learned how to braid rope in scouts.
jaybone@lemmy.zip · 10 pts · 163d
First read this as “it’s braindead.”
irelephant@programming.dev · 5 pts · 163d
Honestly that works as well
mech@feddit.org · 8 pts · 163d
unpopular opinion:
The effort of "tidying up" cables is never worth it.
You'll always end up having to change something and your sophisticated cable management turns a 5 minute job into a 1 hour job, as you open zip ties and dig out cables.
Just use the correct cable length, plug that shit in in a sensible order, label both ends with numbered tape, and let it look messy.
rainwall@piefed.social · 16 pts · 163d
The first step is using correctly sized ethernet cables. Dont use a 12 ft cable for a 3 ft run. There shouldnt be more than a foot of slack in general. You dont need to "tidy" excess you dont have.
Then, for when you do need to tidy cables, use velcro ties, not zip ties. Ideally, you still label all the runs, especially the long ones. Seems a bit redundant for 3ft, but its still helpful.
Marshezezz@lemmy.blahaj.zone · 2 pts · 163d
0x0@infosec.pub · 2 pts · 163d
At least put in the minimum effort of trying to follow the original path (assuming someone worth their salary actually was involved in the making of the site.).
Youd be surprised at the difference in mess from just plugging 'a to b' and 'a to b via c' no matter how many techs having been involved.
mech@feddit.org · 1 pts · 163d
I'd obviously plan the initial setup, start with connecting ports close together via short cables, then put the longer connections on top, avoid unnecessary crossings.
And when switching out a cable, look for the most direct path
0x0@infosec.pub · 3 pts · 163d
Cable trays is the magic word. Any rack larger than 2u that see frequent patching need them no matter what accounting thinks.
Steve@startrek.website · 7 pts · 163d
Uncommon
Newsteinleo@infosec.pub · 7 pts · 163d
God forbid you have to swap out a cable or a port does on the switch. You will spend 20 min trying to figure out what cable goes where
ButteryMonkey@piefed.social · 3 pts · 163d
Or you could mark the bad one, replace the whole set, then unbraid and toss the marked one later in a dumb meeting you have better things to do than attend or something :p
Newsteinleo@infosec.pub · 4 pts · 163d
You still have trace each cable between the patch panel and the switch or thing could end up on the wrong subnet.
markstos@lemmy.world · 6 pts · 163d
Ah yes, a twisted pair.
BatmanAoD@programming.dev · 5 pts · 163d
*sextuple
aeronmelon@lemmy.world · 5 pts · 163d
IT with an E