David Boggs, Co-Inventor of Ethernet, Dies at 71 (nytimes.com) 69
David Boggs, an electrical engineer and computer scientist who helped create Ethernet, the computer networking technology that connects PCs to printers, other devices and the internet in offices and homes, died on Feb. 19 in Palo Alto, Calif. He was 71. From a report: His wife, Marcia Bush, said his death, at Stanford Hospital, was caused by heart failure. In the spring of 1973, just after enrolling as a graduate student at Stanford University, Mr. Boggs began an internship at Xerox PARC, a Silicon Valley research lab that was developing a new kind of personal computer. One afternoon, in the basement of the lab, he noticed another researcher tinkering with a long strand of cable.
The researcher, another new hire named Bob Metcalfe, was exploring ways of sending information to and from the lab's new computer, the Alto. Mr. Metcalfe was trying to send electrical pulses down the cable, and he was struggling to make it work. So Mr. Boggs offered to help. Over the next two years, they designed the first version of Ethernet. "He was the perfect partner for me," Mr. Metcalfe said in an interview. "I was more of a concept artist, and he was a build-the-hardware-in-the-back-room engineer." Many of the key technologies that would be developed over the next two decades as part of the Alto project would come to define the modern computer, including the mouse, the graphical user interface, the word processor and the laser printer, as well as Ethernet.
The researcher, another new hire named Bob Metcalfe, was exploring ways of sending information to and from the lab's new computer, the Alto. Mr. Metcalfe was trying to send electrical pulses down the cable, and he was struggling to make it work. So Mr. Boggs offered to help. Over the next two years, they designed the first version of Ethernet. "He was the perfect partner for me," Mr. Metcalfe said in an interview. "I was more of a concept artist, and he was a build-the-hardware-in-the-back-room engineer." Many of the key technologies that would be developed over the next two decades as part of the Alto project would come to define the modern computer, including the mouse, the graphical user interface, the word processor and the laser printer, as well as Ethernet.
This story makes him sound like a leech (Score:2)
Re: (Score:2)
Is that what you get out of...""I was more of a concept artist, and he was a build-the-hardware-in-the-back-room engineer"? Much like Steve Jobs was the "concept artist" and Steve Wozniak was the "build-the-hardware-in-the-back-room engineer".
Re: (Score:2)
Is that what you get out of...""I was more of a concept artist, and he was a build-the-hardware-in-the-back-room engineer"? Much like Steve Jobs was the "concept artist" and Steve Wozniak was the "build-the-hardware-in-the-back-room engineer".
Up through the Apple 1 and ][, Steve Wozniak was the Conceptualizer AND Hardware AND Software Engineer (along with a little help from Allen Baum); Steve Jobs was the Salesman. Although Jobs is credited with the decision to put 8 expansion slots instead of 4 in the Apple ][ Design ("It'll give 'em something to do") goes the possibly apocryphal quote, the lion's share of the Hardware and Software Concept AND Design of the first two machines goes to Woz.
It wasn't until the Apple /// that Jobs started "Conceptu
Re: (Score:2)
Reading comprehension is hard, I know. Keep trying! You can do it!
Speciality (Score:2)
Metcalfe had the concept on how it would work. Boggs figured out how to implement it successfully. They both get credit because designing something without knowing how to make it work is useless, and you can't build something unless you have a solid design.
A lot more than home offices (Score:2)
Ethernet runs the whole world. Fiber is usually Ethernet. ISPs are connected internally by Ethernet. Wifi is just Ethernet projected into the air-waves instead of on a physical wire. Ethernet is hardly a protocol for connecting home offices and printers. Ethernet runs nearly all the "short hops" on the internet and beyond.
Re: A lot more than home offices (Score:2)
Most (lower-end) printers use USB, making the statement just plain stupid.
Re: (Score:1)
Most (lower-end) printers use USB, making the statement just plain stupid.
Bullshit. I've got over 300 clients and not a single one of them has a USB connection for their printer in use (the usb is there, but it's not used). ALL (as in 100%) use a WiFi printer today. I haven't even seen a non-WiFi home printer in years.
Jesus, my $250 HP Laserjet, that's over 12 years old, has both WiFi and wired ethernet but zero USB ports.
Re: (Score:3)
Re: (Score:3)
By the late 1980s, early1990s, it became very clear Ethernet was awful.
Nearly everything seems awful when vastly more powerful technology becomes available. As you note, the price of switches vs hubs plummeted at some point. When Linksys released the BEFSR41 in 1999, it had a 166 mHz Kendin processor. Now, I'm aware that this was a specialized processor and not intended as a general purpose CPU, and that the BEFSR41 was more than just a switch and that it was somewhat more than what you needed for just a switch. Still, anywhere beyond 4 years before that, if you wanted a proce
Re: (Score:2)
Re: (Score:2)
No, when the 1990s rolled around, everyone suddenly said "Hey, yes, we can finally use Ethernet because previous versions were awful and we'd never have gotten permission to build those out in our buildings!"
My honest feeling is that if we'd stuck with 10base-2, Ethernet would have gone nowhere, even with Novell's support. By the early 1990s, it was still the case that ARCNet and Token Ring were seen as viable alternatives. At least one computer manufacturer (Commodore) threw in its lot with ARCNet in that time period. Networks were for nerds anyway, so who cared?
Well yeah, if we'd stuck with 10base-2, some other system altogether would have come along that took advantage of the newly available technology that enabled faster connections, etc. If there had been no improved version of Ethernet, but a new standard or an improvement of another existing one, we would have gone with that instead.
Classic Ethernet had two problems, the cabling, and the bus contention when you had more than a handful of low traffic nodes on it.
Absolutely no disagreement there.
Which is not to blame anyone. Technology takes time to perfect. But yes, 1980s Ethernet was awful. It contained enough of a vision of what could be to inspire people to fix it, but, like a lot of things people played with in that time period, it wasn't something most people could work with.
I'm just trying to say that awful is relative. Back then we had to configure our own IRQs and DMAs, deal with master/slave configurations for IDE
Re: (Score:2)
Re: (Score:2)
So it's not as if we didn't know things were broken back!
I imagine kids growing up today will be saying the same thing decades from now. Except their complaints will be mostly about software. UIs, DRM issues, etc.
LAN Parties (Score:2)
Ah, LAN parties.
In high school my friend's father was upgrading his company to Ethernet. I bought his dad's old Arcnet gear. Four cards and a passive hub. My father had a large spool of RG-62 lying around the basement so we appropriated it for our network. That started are regular LAN parties.
A few years later we got the itch to upgrade to Ethernet. The cards had finally came down in price that we could afford them. Bought some RG-58 cable, T-connectors and a couple of 50 ohm terminator caps. Eventua
Re: (Score:2)
> At least MAC addresses survived I guess, had Arcnet been adopted by Novell we might still be manually setting addresses for each of our network interfaces...
Ugh, that sounds awful, but I think it also would have meant that IPX/SPX would have been different than they were on Ethernet where clients learned their network address from listening on the wire and their node address was their MAC address.
I still think the IPX/SPX fully qualified network address -- 4 bytes network address learned (or set on se
Re: (Score:2)
Re: (Score:2)
I haven't dug for interviews, but I'd be surprised if IPv4 wasn't just the result of back of the envelope calculations about network scaling from the 1970s.
Re: (Score:2)
Interview: https://dltj.org/article/vint-... [dltj.org]
And 4 billion addresses was more than enough for a TEST network. IPv4 was never meant to be a production network. We're all still participating in the world's largest beta test.
The 1976 paper (Score:2)
Ethernet: Distributed Packet Switching for Local Computer Networks. Robert M. Metcalfe and David R Boggs [archive.org]
Still amazing to think of how much we use today came out of PARC in such a short time:
Back when Jobs first accused Gates of ripping off his idea, the two met in a conference room at Apple, where Jobs ripped into Gates, yelling, "I trusted you, and now you're stealing from us!"
As Walter Isaacson reported in his Steve Jobs biography, Gates responded: "Well, Steve, I think there's more than one way of lookin
Re: (Score:1)
... and true to Xerox's mentality - If it doesn't sell printers what is it good for?
It's one thing to focus your core business - but you never want to be so myopic that you no longer have any vision.
Xerox, because of PARC, could have been Apple.
Re: (Score:2)
Xerox, because of PARC, could have been Apple.
Xerox could have been a lot more than Apple. Xerox could have rivaled IBM at a time when that actually would've meant something. Xerox possibly even could've rivaled Bell.
Re: (Score:2)
The chief problem was timing, as it always is with things. Xerox threw a lot of money into R&D, but as with a lot of R&D, the market isn't quite there yet. There are a whole lot of amazing prototypes built in the 70s and 80s that presaged later developments, and in many cases were the direct inspiration, but a company at the end of the day will still stick to its core competencies, and walk away from what seem at the time the wilder ideas. I'd say the chief issue for PARC workstation always was that
Re: (Score:2)
Xerox could have been Apple, IBM, Microsoft and Sun.
At some point in the past, some Xerox PARC exec was on his deathbed when, in the last few seconds of life, it hit him "Oh, that's what all those things were good for. Damn."
Invented Ethernet? (Score:3)
Re: (Score:2)
It's similar to Alohanet in it's use of bursts of packets and the use of CSMA/CD rather than some synchronous method of media sharing. "Ethernet" includes both the physical media and the OSI data link layer. The original Ethernet developed by Metcafe/Boggs appears to be 10Base5 (coax) based. Needless to say, this has been superseded by somewhat faster and cheaper hardware implementations. I suspect that the packet collision management algorithms of current implementations differ somewhat from the original d
Re: (Score:2)
Better than Thinnet, but needed a hub (Score:2)
Ethernet was better than thinnet because with thinnet if you had a break in your loop everything would stop working.
Thinnet was better than Ethernet because Ethernet requires that everything run back to a hub (back then), which made your cable runs cost more.
Also you could run Token Ring over thinnet.
As an FYI, cable internet is basically a super fast token ring style system because broadcasting.
Re: Better than Thinnet, but needed a hub (Score:2)
Uh, thin net was Ethernet. And you're comparing to Ethernet with more pairs
Re: (Score:2)
No, ThinNet was 10base2. They're talking about the PHY layer, not the signalling layer.
Re: (Score:2)
my mistake, it was all ethernet, just different cable/topology. it was a long time ago!
Re: (Score:1)
Ah... Smok'n Ring... get your CAUs and MAUs - and pray that one of the tiny little tabs on IBMs over-engineered connector the size of a wall-wart didn't break.
Good protocol, though. 16MB TR ~= 100MB Ethernet using hubs. Wasn't until ethernet went to a switched fabric, from broadcast, where its performance started to overtake that of TR.
Regardless - RIP Boggs! Ya done good!
Re: (Score:2)
Hah, I remember an Arcnet installation failing because a terminator with the wrong resistance had been plopped on. Early Ethernet had similar flaws to be sure, but holy smokes was it just a lot easier to work with, and when they came out with RJ-45 cards and switches, I thought I'd died and gone to heaven. Lighter cable, longer transmission lengths.
Re: (Score:2)
Hub? uh no. The first Ethernet I installed, 10-Base-5 (1982) was on plenum-rated Teflon coax with DEC H4000 transceivers to connect DEC VAX 11/780 systems via an AUI cable. It was a bus cable with resistors on each end, there was no hub! Thinnet, 10-Base-2 standard Ethernet (1985) emerged as PC Networking was taking off. You either had a base hub or a bridge that could tie two 10-Base-2 networks together or a 10-Base-5 via an AUI cable. There were also companies that made various repeaters (Fiber Optic, Las
Early ethernet was primitive (Score:2)
I remember helping to install some runs in a computer lab in the 80's. It was a thick, stiff, expensive coaxial cable, twisted pair hadn't been invented yet. None of us had ever done it before so we were struggling to understand what to do.
We were standing on ladders next to the computers attaching vampire clamps to the cable we had threaded up above the ceiling. The clamp had a stiff insulated pin that had to puncture the outside layer of the shielded cable and penetrate in to the central conductor, formin
Re: (Score:2)
One wrong thing anywhere on that cable and the whole network goes down. I don't miss those days.
Re: (Score:2)
As I recall there was a special terminator you had to install on the end of the cable to reduce ringing and noise.
Re: (Score:2)
Re: (Score:2)
Re: (Score:2)
Yes, the teflon cables that we had to use could not be bent very much and so it was a bit of a pain to run it in the ceiling. There were easier cables to use that also used vampire taps but the teflon type was required for fire safety.
There were other things than ethernet, and varieties of eithernet. I remember a long run in a tunnel (used to pass classified docs between buildings) that had two wires on spacers separated by about a foot apart the entire distance of maybe 100+ meters. I was told at the tim
Re: (Score:2)
Maybe it was token ring, that was also a thing back then.
Re: (Score:2)
Didn't the vampire taps have to have a specific spacing on the cable? I think that was part of 10Base5
Re: (Score:2)
Different World Really (Score:2)
Soon there will be no engineer alive that understands how the CSMA/CD works.
I remember sitting in IEEE 802.3 committee meetings where passionate arguments went on about how you couldn't call it "Ethernet" unless it had collisions. That was when they were trying to ratify Gigabit Ethernet and switching technology had progressed to the point that hubs had no cost advantage. But there were die-hards who objected to the change. Good times.
What we call Ethernet today has little resemblance to what Metcal
Re: Different World Really (Score:2)
Does anyone even run network installations that are not full-duplex?
Maybe people that reuse old TV coaxial installs?
Re: (Score:2)
Possibly some legacy network is still running somewhere, but coax-type 10Base adapters will be hard to find.
Re: (Score:2)
Re: (Score:2)
Re: (Score:2)
Yeah, I'm glad the Ethernet II frame format won out against the IEEE frame format with the LLC/SNAP abomination.
Thanks (Score:1)
Didn't even know your name until today but I'm ever grateful for your work! RIP.
"The technology that connects PCs to printers" (Score:2)
Last time I checked, most printers are connected via USB. Secondly, this description of the Ethernet is a F**CKING UNDERSTATEMENT. The Ethernet technology connect pretty much everythng in the f....ing world, dude.
Re: (Score:1)
Last time I checked, most printers are connected via USB.
How long ago did you check? 100% of my 300 clients have a WiFi printer and not a single one of them (NOT ONE), uses the USB port. And that's organic.
Re: (Score:2)
Re: (Score:2)
Depends on if the home has a fixed desktop somewhere. Fair number of households nowadays are laptop-only, or even tablet-only, and no one today wants to have to walk over to the printer and plug in so as to print off a document. I'd be surprised if any printer released in the last decade doesn't have either bluetooth, wifi or ethernet, if not all of the above.
Some consumer routers used to advertise their capability as a USB printer server, but it's been a while since I've seen that in advertising literature
Re: (Score:1)
I'd believe that in an office environment.
Not offices. These are homes.
Everyone in my area has WiFi these days. The adoption rate is been as close to 100% as you can get. The fact is, you speculate and I have direct first-hand knowledge. Nobody is using USB for their printer anymore.
Re: (Score:2)
If this was me (Score:3)
If this was me, at my funeral I'd want:
eth0: Link is down.
What's with this "dies at"? He's already dead. (Score:2)
I've been seeing a lot of headlines that go "[name] dies at [age]". What's with the tense? They aren't going to die in the future. They're already dead.
Not to put too fine a point on it but, this should be "[name] DIED AT [age]". They're dead. That happened in the past.
Re: (Score:2)
Re: (Score:3)
This is true of virtually every headline. "Russia invades Ukraine". "Dam fails". "Man Bites Dog".
Anyone remember Arcnet? (Score:2)
The university I was at bought into ARCNet infrastructure hard. I worked in the new networking group as a Lab to work on the rollout since there were no "networking" classes yet I could take.
We also had a lot of Token Ring loops.
One cannot truly appreciate Ethernet unless they have experienced the alternatives that it replaced.
Re: (Score:2)
I initially had a 4 computer Arcnet network for LAN parties. See my comment here: https://news.slashdot.org/comm... [slashdot.org]
Beyond ethernet - Internet broadcasting (Score:2)
A single computer can do only so much. Thanks, David.