Ethernet is Still Going Strong After 50 Years (ieee.org) 81
The technology has become the standard LAN worldwide. From a report: Ethernet became commercially available in 1980 and quickly grew into the industry LAN standard. To provide computer companies with a framework for the technology, in June 1983 Ethernet was adopted as a standard by the IEEE 802 Local Area Network Standards Committee. Currently, the IEEE 802 family consists of 67 published standards, with 49 projects under development. The committee works with standards agencies worldwide to publish certain IEEE 802 standards as international guidelines.
A plaque recognizing the technology is displayed outside the PARC facility. It reads: "Ethernet wired LAN was invented at Xerox Palo Alto Research Center (PARC) in 1973, inspired by the ALOHAnet packet radio network and the ARPANET. In 1980 Xerox, DEC, and Intel published a specification for 10 Mbps Ethernet over coaxial cable that became the IEEE 802.3-1985 Standard. Later augmented for higher speeds, and twisted-pair, optical, and wireless media, Ethernet became ubiquitous in home, commercial, industrial, and academic settings worldwide."
A plaque recognizing the technology is displayed outside the PARC facility. It reads: "Ethernet wired LAN was invented at Xerox Palo Alto Research Center (PARC) in 1973, inspired by the ALOHAnet packet radio network and the ARPANET. In 1980 Xerox, DEC, and Intel published a specification for 10 Mbps Ethernet over coaxial cable that became the IEEE 802.3-1985 Standard. Later augmented for higher speeds, and twisted-pair, optical, and wireless media, Ethernet became ubiquitous in home, commercial, industrial, and academic settings worldwide."
The real question (Score:4, Funny)
Are you using SNA or DECnet over ethernet?
Re: (Score:2, Insightful)
Re: (Score:2)
I'm still deploying token ring, you insensitive clod (x2)!
Re: (Score:2)
Let me top you.. I'm still using ArcNet (nah, just kidding)
Re: (Score:2)
I am still waiting for the token before I can call someone an insensitive clod again.
Re: (Score:2)
I'm still deploying token ring, you insensitive clod (x2)!
Came here to post the same. I have it deployed over a network of Ubiquiti wireless towers but I am recently having problems with only one of my ~2,780,000 users being able to own the token at a given time as the protocol specifies. Nevertheless, I am working on a patch to mitigate the issue.
Re: (Score:3)
Re: (Score:2)
I used to run IPX over 2.5Mbps ARCnet because it was half the price compared to Ethernet.
Re: (Score:2)
I remember ARCnet. My employer at the time used it because it had twice the range of Ethernet and the buildings were a bit spread out.
Then there was the 10-base-2, over coax, also with twice the range of twisted pair. Don't forget the terminators.
Re: (Score:2)
ARCnet was used extensively because it was cheaper than Ethernet. In fact, at the time, it was actually hard to tell who might win - ARCnet or Ethernet, because ARCnet was popular as a LAN technology due to its extremely low price and low overhead, so by the time Ethernet came out, ARCnet was widely deployed.
This video goes over ARCnet and then explains why Ethernet won out i
Re: (Score:2)
Re: (Score:2)
My school's large computer lab filled with IBM PS/2 Model 25 (8086 all-in-one) all ran ARCnet and logins and storage were managed by NetWare. The reason I got into ARCnet is because when machines in the lab were retired, due to failing floppy drives that were overly expensive to replace. The English teacher (who also doubled as the school's sysadmin) gave me some of the spare cards for my own use. (!)
So I got into ARCnet roughly when Ethernet cards were getting cheap. At home I ended up with a mixed setup o
Re: (Score:2)
Ah, Satan's Network Architecture. There are vestiges of that in the TN3270 proocol.
Re: (Score:2)
Even Satan would be left wondering where all his money was added a visit from IBM's consultants.
Still Less Complicated Than My Last Relationship (Score:5, Insightful)
Ethernet's like that Nokia 3310: indestructible. And let's be honest, there's something deeply satisfying about that 'click' when you plug in an Ethernet cable. It's the sound of stability in an unstable world. Ethernet, the unsung hero of LAN parties and late-night coding sessions.
Re: (Score:3)
What's interesting to me is that ethernet continues to be more reliable than wifi even though wifi has improved incredibly in recent years. Personally, I would choose 100mb ethernet over 200mb wifi. Part of this is due to latency issues or external interference, problems intrinsic with wireless but not with wired connections. The biggest difference between the two, in my opinion, is in how the operating system handles it. If my Internet connection has difficulties when I'm connected via ethernet, no problem
Re:Still Less Complicated Than My Last Relationshi (Score:4, Interesting)
Growing pains
I remember when we went from 10base2 (thinnet) to 10base-t in the early 90's
The runs from workstation to wiring closet stayed the same, but the 10base-t that was installed could not handle the long runs that the thinnet handled
As are result the ethernet became VERY unstable and large file moves from workstation to server became impossible
They eventually figured it out and added more wiring closets to support shorter runs from switch to switch and switch to workstation, but it was a real head scratcher there as the networking people were determined to move to the new standard and did not realize that the runs exceeded the product capability
Re: (Score:2)
You actually had cable runs longer than 100m? I realise 10base2 can nominally support 185m segments, but you'd be really pushing it if that was anything besides a link between two bridges or something.
Re: (Score:2)
I am not 100% on this, but I think that prior CAT versions had shorter runs
In any case the architects had designed for thinnet and the earlier versions of 10Base-T were shorter
Re: Still Less Complicated Than My Last Relationsh (Score:2)
The original 10baseT with Cat3 cabling supported a maximum cable length of 100m. I realise it's substantively less than the 185m supported by 10base2 with RG58A/U coax, but in practice I never actually see this cause an issue when upgrades were carried out.
For very small setups where everyone was on a single 10base2 segment, you'd put the hub in the centre, and that would guarantee the most distant nodes would be no more than 100m away. For larger networks, the segments connecting groups of nodes were never
Re: (Score:2)
In the case that I am referring to
1. It was a new building, for a large agency, with a dedicated server room and the rest a cube-farm
2. The CIO HATED ceiling drops and would not allow for them. In wall (networking closet to server room) and in floor (cube farm to networking closet) conduit was used throughout
3. Thinnet was the agency standard when the request was put out for bid
4. CIO decided to wholesale upgrade to 10Base-T and would allow no Thinnet in the building
5. Hilarity ensued as my GIS team tried t
Re: (Score:2)
It was all hubs not switches back then. I thought those used dumb broadcasting. Did that affect reliability of one of the cables was bad?
Re: (Score:2)
Hubs are just extensions of the existing lines, switches allow for error correction and re-transmission, thereby reaching longer distances
Increasing distance [automation.com]
Twisted-pair segment lengths cannot exceed 100 m and, therefore, the need to extend network distances is frequently encountered. If there is a need to extend an Ethernet network, a switch can provide an advantage over a repeating hub.
Re: (Score:2)
OK yeah that makes sense. I never had much exposure to hubs. When twisted pair started really displacing coax, it was for me a free source of networking hardware that I could not easily afford otherwise. I did the odd LAN party with coax. I certainly used base-T stuff but by the time I was in a place where I might want my own switch or hub (rather than simply plugging into a wall), switches had come down in price enough that I got switches.
I don't actually recall if I ever personally used a hub in a non dem
Re: (Score:2)
And yes, Ethernet switches even existed in the early 90's [wikipedia.org], but most people did not want to pay the premium for them over hubs
Re: (Score:2)
Re:Still Less Complicated Than My Last Relationshi (Score:5, Informative)
RF is way, way more problematic than people would like you to believe. I used to work in a lab simulating RF propagation issues over a significant land area. It was actually kind of challenging. We could simulate terrain, weather effects, etc. Antennas, transmit power, frequency all impact things but the bottom line is that getting a signal from point A to point B at a particular minimum strength in anything but perfect LOS is not guaranteed.
Re: (Score:3)
That's why error detection/correction, resend, frequency hopping, quality/latency and congestion detection and more are part of the science. If it was just packaging and acknowledgement, we'd be done long ago. And to be clear: wired transmission has similar issues, down to the circuit level.
When one thinks of those ubiquitous bluetooth earbuds that are connected to each other and to a phone, inside a wifi home (esp. if using starlink), then the number of layers involved in delivering high-fidelity soun
Re: (Score:2)
What's interesting to me is that ethernet continues to be more reliable than wifi even though wifi has improved incredibly in recent years.
WiFi as a specification may have been improving but support of much of the newer features still lags.
I'm seeing brand new, supposedly "high end", Wi-Fi hardware still being stuck on 2.4 GHz. I'll see things start acting funny because I'm warming up something in the microwave, something I thought was left to history 10 or 15 years ago. It's not like I can just get a $30 USB adapter for what is missing support for WiFi on some band other than 2.4 GHz because the devices lacking support are things like my TV
Re: (Score:2)
This sort of thing surprises me when I hear of things like mobile phone hotspots. I think to myself that they must be incredibly slow, because the mobile 4g speeds are amazingly slow. But it turns out that the slowdown is mostly due to the phone; it's busy wasting most of it's processor cores in prefetching data to show you more ads, and wasting the bandwidth on bulky ads, and the small amount of data I really wanted is delayed no end. For sure, wifi on the phone is astoundingy faster than cellular data e
Re: (Score:3)
If I had a dime for every time someone was having a connectivity issue while in the office and I told them to make sure their network cable was plugged and received, "What's a network cable?", I'd have enough money to buy a rib eye steak.
When I tell people that their wireless connection still uses cables, that blank look on their face is precious.
The vast majority of people don't understand how ubiquitous ethernet is and how its use keeps the modern life going.
Re: (Score:2)
The state put out a decent Cybersecurity Awareness course video that we encouraged our users to watch. It explained the basics of what to watch out for, what not to do, etc... The video suggests unplugging the network cable first if you suspect you've been infected and shows you how. Almost immediately calls started coming in from users saying "I looked behind my PC and I don't have a blue cable".
Re: (Score:2)
This sounds more like a configuration problem in your office. Why should the network cable need to be plugged in if you also have a wireless coverage? If I pull out my network cable right now, nothing will happen and everything will continue working just fine, wirelessly.
Maybe gaming will get a bit more latency and the occasional dropped packet, but otherwise everything will continue to work.
Anyway excuse me while I need to duck out. We can continue this conversation though as my phone transparently moves f
Re: (Score:2)
It was just fantastic from the off even with the 50 OHM filter nightmare.
Loved it.
Re: Still Less Complicated Than My Last Relationsh (Score:2)
Re: (Score:2)
Re: Still Less Complicated Than My Last Relations (Score:2)
Re: (Score:3)
To be a nerd...it was a backwater posting until the discovery of the wormhole in episode 1, and by that time Sisko had a personal relation with the weird aliens and they asked to keep him. Anyway. this is standard TV stuff, not a case of Starfleet doing racially based promotions.
Sisko didn't leave his wife and son forever, he came back (the actor playing Sisko insisted on this one because it seemed racist.)
Re: (Score:2)
And of course they made Sisko the "angry" captain.
This was because he was the first captain not created by Rodenberry - Rodenberry who thought good scifi was having future humaity be flawless.
Re: (Score:3)
Re: (Score:2)
Of course, Sisko not a captain in the second sense until they had him start running the Defiant. He commanded a base, not a ship.
Re:lol (Score:5, Informative)
The hilarious part about DS9 was the casual racism that snuck in, like, they made Sisko start out as a Commander for no reason whatsoever.
Do you know how the actual Navy works? Your rank, generally speaking, comes down to how many people are under your command. Most ships in the USN are commanded by -- wait for it -- Commanders. If you're in command of a submarine or destroyer, it would be rare to be a Captain, unless you were also a squadron commander, and most of those end up with postings shoreside. When the USN had much smaller ships, like submarine chasers, the "Captain" was most likely a Lieutenant, perhaps even a Lieutenant Junior Grade. JFK was a LTJG when he commanded PT-109.
The fuck-up here is later versions of Star Trek thinking that someone like Janeway -- whose ship had a crew of ~150 people -- would be a Captain. She'd be a Commander. When she earned the rank of Captain, she'd lose Voyager and get a much bigger command, perhaps even a boring ass one on a Starbase somewhere. You don't get to take a promotion in the military and keep your small cozy command just because you happen to like it.
Sisko being a Commander of what was SUPPOSED to be a backwater unimportant posting with a crew smaller than Picard had was absolutely correct. When the posting later became hugely important, and he was on the front lines of a brewing war, they promoted him to Captain. They also gave him a lot more people. We never learned most of their names (unless the script called for a tragic death) but if you look at the background actors over the years in DS9, there were a lot more Starfleet types running around after Sisko's promotion.
The real fuck up in DS9 came during the war arc, when Captain Sisko was suddenly in charge of a fleet of 600+ ships. That'd be well over 100k personnel. A command like that in the USN would only go to a Full Admiral. If they had wanted to maintain accuracy, Admiral Ross would have been in command of that fleet, with Sisko at his side, Sisko being the Subject Matter Expert in the Dominion. Ross would handle the strategy and Sisko would handle the tactics.
If they had wanted to really maintain accuracy, Sisko and his crew would not have been on the frontline at all. They'd be back on Earth, at Starfleet Academy, training new officers and crew members. You don't leave your most experienced people on the frontlines. You bring them home and have them share their hard won knowledge as broadly as possible.
Re: (Score:3)
Can't comment on the military stuff in your post, but you left out the elephant in the room - did DS9 still use Ethernet? Or did the Bajorans have their own, incompatible LAN standard?
Re: (Score:3)
DS9 was Cardassian tech, so it probably used IPX, and Chief O’Brien had to conjure up a bridge over a cup of raktajino.
Re: (Score:2)
Re: Still Less Complicated Than My Last Relationsh (Score:2)
Psst WiFi is Ethernet.
Virtually everyone makes this mistake, so don't feel bad. But it IS Ethernet. If it weren't you would need to translate or encapsulate Ethernet in order to connect to the Internet with WiFi. WiFi devices have MAC addresses for a reason...
Re: (Score:2)
No it isn't. WiFi uses a completely different protocol.. It s very Ethernet-like, but the packet format is not Ethernet. (It's this difference that lead to people breaking WEP).
And you do need to translate between Ethernet and WiFi - that's the job of the Access Point. Its
wifi does not have POE and WIFI AP's need wired ne (Score:2, Flamebait)
wifi does not have POE and WIFI AP's need wired network to get good speeds back to backend network.
the newer wifi do have better speeds but the range is going down.
Re: (Score:2)
But if you're going to allow WiFi to be new equipment, it needs to be compared to gig or 10 gig ethernet.
But what is left of the original? (Score:3)
Re: (Score:3)
Re: (Score:2)
Ethernet 2 frames differ from the original Ethernet frames, I believe.
Re: (Score:1)
All the layers haven changed except the heart of ethernet: MAC addresses, ethertype/llc and CRC. Everything above that is another protocol (upon protocol, upon protocol, upon protocol). The layer below we still generally call ethernet, even though there's not much in common between the old twisted pairs, 802.11 wireless, and optical shenanigans over fiber(s).
Re: (Score:2)
Re: (Score:2)
TLS and http3 are a layer above, as are CGNAT and ipv6.
Technically, the cables are a layer below, but I'll grant most people think of the physical and link layer as one. But for all of the changes to the cables, you can STILL connect a thick ethernet up to a larger twisted pair network using the right media converter.
Re: (Score:1)
The OSI model is a concept that doesn't always map well to the real world. I sometimes wonder why the OSI model is taught, or at least when taught that there isn't more emphasis on this being a concept to aid in understanding the function than something existing in reality. Separating the layers conceptually might be helpful in troubleshooting, programming, or something but can breakdown real fast in the real world. Sometimes is it difficult to define where the layers of the OSI model map to the real wor
Re: (Score:2)
True enough. Mostly I just characterize things as physical, layer 2, or layer 3. UDP and TCP may technically be layers on top of IP, but they feel more like flavors than layers. Above that is just "stuff sent over the network".
This proves one thing (Score:3)
Resilience and easy configuration beats performance
Other tech like token ring offered higher performance under congestion, but required complicated setup
Ethernet tolerated congestion and collisions by simply waiting and trying again
Ethernet cost less and when switches came TR died (Score:2)
Ethernet cost less and when switches came TR died.
Re:Ethernet cost less and when switches came TR di (Score:4, Funny)
also, just try losing your token.
Re: (Score:2)
Switched token-ring over UTP was a thing for a while back in the early '90s. I set up a small 16Mbps TR LAN (20 nodes) running on Cat 3 (telco cable) back in those days to support a start-up. We chose that over Ethernet because 16Mbps > 10Mbps Ethernet and, at the time, the TR switches were cheaper than Ethernet switches. Eventually we migrated to switched 100Mbps Ethernet.
Re: (Score:2)
Resilience goes to Ethernet though. No need to search the carpet when the token falls out of the ring...
Original Paper .. (Score:4, Informative)
Abstract: “Ethernet is a branching broadcast communication system for carrying digital data packets among locally distributed computing stations. The packet transport mechanism provided by Ethernet has been used to build systems which can be viewed as either local computer networks or loosely coupled multiprocessors.”
Vampire Tap (Score:2)
I remember installing RG-8 10 Mbps Ethernet coaxial cable and placing external transceivers at the marked vampire tap lines on the cable. Crap, I'm getting old.
Re: (Score:2)
Yeah, I remember having to actually drill into that fat-ass cable during the tap installation(s). I think that the taps came with a special bit, or tool, or something.
Seemed like a strangely atavistic way to add a network node.
Robert Metcalfe inventing/selling ethernet (Score:3, Insightful)
I remember reading somewhere a comment by Robert Metcalfe that he didn't get rich by inventing ethernet, he got rich by selling ethernet.
Is this the one with apps in it? (Score:2)
Showing cracks (Score:2)
It's been showing cracks now for years, since the invention of things like Infiniband. It's going strong and won't be replaced, but for things like data-intensive, low-latency workloads it's been eclipsed by other technologies. This is more true in hyper-converged and cloud environments.
Ethernet is cheap and adequate (Score:2)
But it's slow compared to other network protocols, with a latency usually measured in milliseconds.
Its main strengths are being a public standard, easy to implement, and in consequence extremely cheap. It's also very flexible. And, for most purposes, it is perfectly adequate.
Infiniband's latency is between 0.5 and 1.5 microseconds - thousands to tens of thousands of times smaller than ethernet. And infiniband is faster, maxing out at 600 gigabits per second, compared to the fastest ethernet cards that top o
Re: (Score:2)
Infiniband hardware is really cheap on the second hand market. Switches can be an issue as some of them will only do tcpip as a licensed option.
Re: (Score:2)
Yeah, my home lan is infiniband. Quad, so it tops out at 40gbps. The cards were about the same price as the 10gbps ethernet cards I also bought. But Ebay has a limited supply and they're not under any kind of warranty. Still, it's fine for home use.
Re: (Score:2)
I would use more Infiniband if the power draw wasn't so damn high.
A gigabit ethernet switch can comfortably live inside 3W.
Competition, I guess, but my 90's 10Base-T hubs were cool too.
I miss the old vampire taps, but I like speed (Score:2)
At another job we were upgrading a PC to include thin Ethernet. But, we did not have a full length loop of coax cable. So we did something that should work to verify software functionality. This was using a BNC T and sending the loose end to the new PC. This was done in the e
Re: (Score:2)
Ethernet != 802.3 (Score:2)
Allow me to be a grumpy pedant for a second (having spent over a year of my life arguing with network stack teams about this). Ethernet and 802.2/802.3 are not the same things. They have different header fields. They use the same physical layer(s), at least back in the 10BaseT days.
No one but a network nerd will know or care about the differences. From a practical perspective, the only way this was ever visible is the MDU of 802.2/802.3/SNAP was three bytes shorter than Ethernet and if you configured your n
So much better than it's predecesor (Score:2)
I was lucky (?) enough to spend a fair amount of time working with Chaosnet, That Which Came Before/Around The Same Time. It used 1/2" CATV cables and required you do make a vampire tap using specialized tools in specific locations on the cable if you wanted to add a port. Get the depth of the coring operation on the side of the cable wrong, and you just screwed up an entire cable run, which would need to be replaced. Ethernet cabling with RF connectors was SO much nicer to deal with!
Ethernet. (Score:2)
10GBase-SFI <full-duplex,rxpause,txpause> (Score:2)
Sort of true, but misleading. (Score:1)