Once you get over the copious standardese, and the fact that the standards describe some things that seem like a good idea yet have never been implemented (e.g. 100BASE-T2), they are a good source of very detailed and authoritative information.
Is that 100 Mbps Ethernet vs telephone pin assignment correct? I thought that the inner pairs that are treated like RJ11 would be the telephone lines and the outer pairs would be Ethernet.
Edit: apparently it is right. I hate how intuition based on principles of simplicity and signal integrity are betrayed at every level of classic Ethernet.
RJ45 8P8C is compatible with telephone RJ11 6P6C for two lines. The inner pins are one line, the next outer pins are second line. The third line in 6P6C doesn't work because the third and fourth pairs are together. But RJ11 using more than two lines, or even multiple lines, are pretty rare.
Right but all that means is that you have a termination that can be implemented in cases where not all features are available. It indicates that it's either the wrong termination or the wrong jack.
Tangentially related question: Does anybody have a good overview resource how to configure networking in a linux system (exluding graphical tools)?
I know there are multiple ways to do similar things and settings are somewhat scattered into different files, but this is why I wondered if there is some kind of overview that I could use as a starting point.
Networking for System Administrators is often my recommendation for this. The author walks all the way up the stack from the physical layer up and name-drops every program and file he uses: `ethtool`, `/etc/hosts`, `/etc/resolv.conf`, `ipconfig`, `netcat`, `tcpdump` ...
I quite liked the accessible style, and got a lot out of making Anki flashcards for all of them.
Hi! If you don't mind, I would love to see that Anki deck. I'm a heavy Anki user myself, too. My gmail username is the same as my HN username. Thank you!
Excellent, I'll email you later on tonight. It'll be a good excuse to give the cards a once-over for formatting, etc and make sure they're all up to snuff!
The "different files" you refer to are likely just different distros ways of configuring scripts that use netlink/ip. Each distro will do something differ, or use something like systemd-networkd/networkmanager. There's no one way.
netlink is the API, and `ip` is the canonical CLI tool for interacting with the tool.
I found that nothing made sense until I learned the basics, which I did by buying the CCNA books. They start at a very low level, explaining things like what an IP address is, what a network mask does, and so on.
After you know that sort of thing, all the things built on top make a lot more sense.
SFP+ got much cheaper now, so you can get 10 Gbps way easier than in the past.
The issue seems to be switching. To switch so much data you need either a powerful CPU or some kind of ASIC and apparently all those ASICs are proprietary and no one wants to make 10 Gbps switches with open firmware becasue of that.
You can get some relatively affordable 10G-BaseT SFP+ transceivers these days too. My home server has a DAC to a switch whilst my desktop is connected at 10G over Cat6. I don't really have the ability to saturate 10G but can quite easily exceed 1G for file transfers. It would be nice to get a laptop dock with 2.5G+ Ethernet as well, but I haven't gotten that far yet...
Yeah, haven't tested it on Windows, but they supposedly support it too.
I was also looking for some good passively cooled switches for SFP+ and they are hard to find. Mikrotik CRS305-1G-4S+IN looked good, but I'd prefer something with open OS on it. They don't seem to exist.
Wish it talked about the cursed adapter. Would you have two of these adapters on each end, and the single cable would be split into two separate pairs for 100 Mbps ea?
The reviews of them are full of negative comments from people who don’t understand what they are, and think they’ll split one network cable into two, like a three-port switch - even though most of the listings are quite clear that you need to connect two ports at each end of the connection.
Most of the listings are, though, quite evasive about the fact that, even used correctly, they’ll not provide throughput over 100Mbps.
That's the basic idea, yes. The link in the caption [https://images-na.ssl-images-amazon.com/images/I/61llDf3WTvL...] is unusually helpful, showing each port using four connectors of the uplink. So: 100mbps max, and you'll just get garbage on the other end unless you're careful.
It's an ugly hack, but hey, there's a lot uglier in active use.
That sounds like a clever workaround, utilizing multiple connectors for each port to maximize efficiency. But I'm curious, what specific challenges arise when dealing with the potential for receiving garbage on the other end, despite being careful?
I use Ethernet everywhere I can and only use wifi if I need to. When you have dozens of devices on a network, reducing the WiFi congestion as much as possible helps all devices. I also use it as a backhaul between mesh wifi access points, so even in an all wifi setup, there is still a good use for direct wiring.
I work from home so I'm connected 8-10hrs a day. Ethernet has always been more reliable and faster. It's one less complication and in urban areas a significant one.
I have lots of devices on WiFi but my main computers (work and gaming :) are wired
If you have the cabling to dedicated locations where computing happens. Be that desk or desktop or tv or such it makes lot of sense. Bandwidth is superior, reliability is superior. And it leaves the air to all other devices. Which in certain locations can be quite a lot.
Both of them together is often best of the both worlds. Connect desktops, NAS and other such devices you do not move with ethernet and then use WiFi for devices being moved around.
The other reason is bandwidth. WiFi can't compete here either: if you have an office full of computers, running separate CAT5e/CAT6 cables to each computer will result in FAR higher bandwidth than if those computers are all sharing the same WiFi.
I only want mobile items on WiFi. Anything that stays in one place I have wired up. Running my own Cat6 cables throughout my house, very easy to make your own cables. Also means my mesh router has a wired connection to the first router.
Wi-Fi is Ethernet, just with a different physical layer (Layer 1). Layer 2 is slightly different but still contains all of the Ethernet hallmarks (such as MAC addresses, ARP, etc).
Fiber is also almost always Ethernet these days, as well.
My thoughts exactly. I could have used this article when I was researching to wire my whole house for CAT5 25 years ago. I moved house last month and put in a mesh network. It felt far too easy!
Yes. In 568B pin 1 would be Orange-Stripe (left side looking from top of plug). Very odd indeed. This pinout has been burned into my brain since high school - and I've never seen it represented backwards as their diagrams show. I even have the printed 802.3 spec from when I worked for a defense contractor a buddy of mine gave me as a gag gift. I've referenced it over the years more than expected.
But 568A and 568B are specifically color coded to the pins. And these diagrams are wrong.
Thanks for the confirmation. I spent way more time than I care to admit double checking this. I tend to not make my own cables anymore and just use punch down sockets, so it’s been a while.
> It used to be that you could use a coaxial cables for Ethernet and wire a single cable between computers, but those days are long gone, and the speeds were low enough that the major reason to miss those times is nostalgia.
ah yes. I tried to make drama out of Xerox's whole Ethernet going down, because of a storm just like we're having now (except no lightning).
========= "Lightning Strikes" chapter of Inventing the Future =====
Thursday, January 5, 1978. Janet was visiting Palo Alto again, something she never tired of. Today, she was just supposed to check in with people and learn whatever she could about the Pilot operating system, the product hardware, Ethernet, user interface, and all that.
In one of those magic changes of name that Xerox never officially announced because that would let an enemy keep track of code names, the product was now called “Star,” She hoped she wouldn’t slip up today and call it “Janus.”
This time she took an earlier flight than she had been lately, the 7:00 am PSA flight out of LAX. She hated having to get up at an obscene hour, but really, if you took a more reasonable flight, you got to the office at 10:45 or later, and that made it a pretty short day. This morning, the roads were all wet and littered with tree branches and leaves. Apparently, it had stormed the night before.
She first went to Grant’s office to talk about the latest marching orders from corporate management. Grant was pacing back and forth, frowning, holding the base of the phone in his right hand and the receiver in his left. He glanced at Janet, smiled briefly, and motioned to the guest chair.
“How much of the network is down?” The answer seemed to make him even more worried. “Does Michael know? Okay, I’ll tell him.”
“Who’s looking into this?” He wrote down some names on a yellow pad.
“Um, yeah. We have Peter McColough and Bill Glavin flying out tonight for a demo.” He stared at Janet, knowing he just dropped the name of the Chairman of the Board. Janet knew McColough was adamant that no exec would ever touch a keyboard—that was for their secretaries. The idea of an expensive computer like Star for knowledge workers, whatever that meant, was dubious to him. His rule of thumb was that you shouldn’t buy a piece of equipment for a secretary that cost more than her annual salary. The execs were already a hostile audience. A bad demo was about the last thing Grant needed.
“I’ll let you get back to it. Keep me posted.” He hung up and turned to Janet, “The entire Ethernet is down. Excuse me, but I need to work this out.” He started for the door, then paused. “On the other hand, do you want to help out?” Grant strode briskly toward Michael’s office with Janet trying to keep up.
Once you get over the copious standardese, and the fact that the standards describe some things that seem like a good idea yet have never been implemented (e.g. 100BASE-T2), they are a good source of very detailed and authoritative information.