dslreports logo
 story category
Ethernet Will Never Work
1974 Xerox PARC memo....
Blogger Bill Heyman offers up this amusing original memo from Xerox PARC (Palo Alto Research Center) employee Robert Bachrach to Bob Metcalf and Dave Boggs suggesting that Ethernet will just never work. The discussion over at Reddit notes that Bachrach wasn't entirely wrong in his analysis, though. They also note that Bob Metcalf has had more than his fair share of innaccurate predictions.
view:
topics flat nest 

N10Cities
Premium Member
join:2002-05-07
0000000
Asus RT-AC87

2 edits

N10Cities

Premium Member

Hmmmm......

If we were still using the old bus topology coax (which the memo points out), then he might have a point! One cable cut and the whole shebang is gone....(or did that only happen with Token Ring?)

ropeguru
Premium Member
join:2001-01-25
Mechanicsville, VA

ropeguru

Premium Member

Re: Hmmmm......

Old bus topology would have gone down as well. As far as I can remember it required a terminator on each end.

phxmark
What Country Are We Living In?
join:2000-12-27
Glendale, AZ

phxmark to N10Cities

Member

to N10Cities
said by N10Cities:

If we were still using the old bus topology coax (which the memo points out), then he might have a point! One cable cut and the whole shebang is gone....(or did that only happen with Token Ring?)
Token ring would only die on a client cable or feed cable cut with the old STP BIG BLACK CONNECTORS using a passive MAU. The old shorting type connectors so that when it was unplugged the token would still get passed, but a cut was usually fatal to the ring.
Time4aNAP
Premium Member
join:2007-04-09
Des Plaines, IL

Time4aNAP

Premium Member

Re: Hmmmm......

said by phxmark:

Token ring would only die on a client cable or feed cable cut with the old STP BIG BLACK CONNECTORS using a passive MAU. The old shorting type connectors so that when it was unplugged the token would still get passed, but a cut was usually fatal to the ring.
IME in the one token ring shop that I worked in, the weak point was at the NIC. Every packet had to pass through every NIC along the way. While the big square wall sockets were shorting, the DIX connectors at the NIC end were not. Even when the wiring was perfect, all it took was a single NIC's transceiver to fail to cause a similar (and harder to find) outage. And they did fail, too often for my tastes. IIRC that was still a weakness with the 16 Mbps UTP topology.
Time4aNAP

Time4aNAP to N10Cities

Premium Member

to N10Cities
I worked for an outfit that was deploying thinnet in the early 90s. It had all kinds of weak points. The off-the-shelf BNC "T" connectors used at each NIC led to lots of inadvertent disconnections. You didn't need a cut, a good tight kink in the coax was enough to compromise the transmission line. But it did save a bundle on hubs and star wiring installation costs.

Maxo
Your tax dollars at work.
Premium Member
join:2002-11-04
Tallahassee, FL

Maxo

Premium Member

So what?

The winds of technological change are too unpredictable. I take little stock in what the experts think will and will not work.
We all know Bill Gates original opinion on the internet, yet who can deny he has made some sort of impact on the internet. A few businesses use their web server, and more than a small handful are using their OS and browser to view those pages.

Mike B
@rr.com

Mike B

Anon

Xerox also...

Xerox also thought the mouse was a joke and destined to fail.
Time4aNAP
Premium Member
join:2007-04-09
Des Plaines, IL

Time4aNAP

Premium Member

Re: Xerox also...

For a company that had the foresight to realize that they had to evolve beyond being just a copier company, Xerox did an amazing job of neglecting the technology that could have turned them into the Xerox of computing.
scooper
join:2000-07-11
Kansas City, KS

scooper

Member

Us old timers have seen it all ...

I've been around networking long enough that my HOME network was the coax variety of ethernet (10B2)! I've also worked with Token Ring, as well as UTP.
dave
Premium Member
join:2000-05-04
not in ohio

dave

Premium Member

Re: Us old timers have seen it all ...

Which coax?

The thick one with the vampire taps, or the daisy-chained cheapernet?

DHRacer
Tech Monkey
join:2000-10-10
Lake Arrowhead, CA

DHRacer

Member

Re: Us old timers have seen it all ...

He says 10B2, which was the Thinnet which used BNC connectors prevailingly.

It was 10B5 that used the thick cables that required vampire taps.

MrMoody
Free range slave
Premium Member
join:2002-09-03
Smithfield, NC

MrMoody

Premium Member

Re: Us old timers have seen it all ...

Ah, the Yellow Garden Hose! I've only ever seen that at a university. I had thinnet myself though, hooked up to play multiplayer Doom II ... fun days.
dave
Premium Member
join:2000-05-04
not in ohio

dave to DHRacer

Premium Member

to DHRacer
Ah, right. I didn't recognize the abbreviation (at DEC, we always spelled it out: 10base2).
Expand your moderator at work
scooper
join:2000-07-11
Kansas City, KS

scooper to dave

Member

to dave
Yes - my first home network used 10B2 - at work, my first exposure was to the Vampire taps
Time4aNAP
Premium Member
join:2007-04-09
Des Plaines, IL

Time4aNAP to scooper

Premium Member

to scooper
Seen ARCNET? SNA?
dave
Premium Member
join:2000-05-04
not in ohio

1 edit

dave

Premium Member

Re: Us old timers have seen it all ...

said by Time4aNAP:

SNA?
Well, as it happens, I was a member of the software team that built various DECnet/SNA Gateway products...

(And IBM channel cables make original ethernet wires look positively skinny. I like to bore people with tales of taking a hacksaw to a VAX in order to get an IBM channel interface installed...)
Time4aNAP
Premium Member
join:2007-04-09
Des Plaines, IL

Time4aNAP

Premium Member

Re: Us old timers have seen it all ...

LOL They didn't call IBM "Big Blue" for nothing!
dave
Premium Member
join:2000-05-04
not in ohio

dave

Premium Member

Yeah, but, no, but ...

It's fine to find an example of a successful technology, and then have a laugh at the old memo that said "this can never work".

But what about the no-doubt numerous examples of failed technologies, and the old memos that said "this can never work"? Why no news postings on how great the memo writers were?

(But it is a pretty amusing memo, now we have the benefit of hindsight).

The viewpoint in that memo, that you can't run a reliable network without solid guarantees about the maximum time to getting a transmit slot, etc., was pretty common at the time. It was the official viewpoint of IBM (for obvious reasons), to give but one example.

Of course, the Ethernet we run today is nothing like the Ethernet being criticized in that memo. As Metcalfe once said, Ethernet will be around for ever, because whatever replacement technology gets invented, it'll be called "Ethernet".

PolarBear03
The bear formerly known as aaron8301
Premium Member
join:2005-01-03

PolarBear03

Premium Member

Re: Yeah, but, no, but ...

said by dave:

But what about the no-doubt numerous examples of failed technologies, and the old memos that said "this can never work"? Why no news postings on how great the memo writers were?
Same reason the 6 o'clock news only reports tragic, sad stories of bad things, never good news - it's more entertaining.
Time4aNAP
Premium Member
join:2007-04-09
Des Plaines, IL

Time4aNAP to dave

Premium Member

to dave
said by dave:

(But it is a pretty amusing memo, now we have the benefit of hindsight).
Hindsight is one thing. What I found amusing were references to statistics that Mr. Bachrach used to bash the proposal. This guy had to have been a bean counter who truly believed that he knew more about engineering than engineers.

This one is the capper: "The fallacy in your conception is that the stations should be transmitting randomly." Remember that this memo is dated after Vint Cerf had set up shop at Stanford, and had made great progress on TCP/IP. By 1974, the writing was on the wall that Bellcore technology was not the future. Mr. Bachrach may have been the only one in Palo Alto who was oblivious to this.

phoneboy2
@shawcable.net

phoneboy2

Anon

Token ring was better before layer 2 switches

Token Ring was better in theory and was more popular for a time in the 90's with enterprises. I remember trying to decided whether to go Token Ring or Ethernet but Ethernet was much cheaper and more practical for small networks using hubs.

Then reasonably priced layer 2 switches came out which effectively eliminated all Ethernet short comings and Token Ring seemed to dissappear overnight.

Kfedka
Premium Member
join:2005-05-06
Spokane, WA

Kfedka

Premium Member

Re: Token ring was better before layer 2 switches

Dang! You guys are old school!

PolarBear03
The bear formerly known as aaron8301
Premium Member
join:2005-01-03

PolarBear03

Premium Member

Re: Token ring was better before layer 2 switches

For sure!

calvoiper
join:2003-03-31
Belvedere Tiburon, CA

calvoiper

Member

Re: Token ring was better before layer 2 switches

Switching, cabling, switching, cabling....

One gets cheaper and forces improvements in the other. Then the cycle reverses.

So pass the sands of time....

calvoiper
dave
Premium Member
join:2000-05-04
not in ohio

dave to Kfedka

Premium Member

to Kfedka
Well, before Ethernet, we used to build networks from point-to-point links. The really fast machines used to have megabit-per-second twinax connections, though 56K was more common.
Time4aNAP
Premium Member
join:2007-04-09
Des Plaines, IL

Time4aNAP

Premium Member

Re: Token ring was better before layer 2 switches

said by dave:

Well, before Ethernet, we used to build networks from point-to-point links. The really fast machines used to have megabit-per-second twinax connections, though 56K was more common.
I'm surprised at how many new EIA-422 networks are coming out. I have brand-new pro audio equipment that uses it!

n2jtx
join:2001-01-13
Glen Head, NY

n2jtx to phoneboy2

Member

to phoneboy2
said by phoneboy2 :

Token Ring was better in theory and was more popular for a time in the 90's with enterprises.
My previous employer was in the fortune 10 and everything was Token Ring except for two corporate divisions, mine and one other. That was due to the fact that we were the only ones running DEC hardware at the time. It was a very tough internal struggle keeping TR out of the organization. It cost three times as much for a TR card versus an Ethernet card and when they started getting integrated onto the the motherboards of our PC's the price was nothing versus $300. We managed to get some traffic between our network and the corporate network using an IBM 8209 bridge. However, once corporate decided the future direction was going to be TCP/IP, I migrated my entire Ethernet network over to TCP/IP. Once corporate starting migrating the mainframes over, the Ethernet isolation disappeared. After I left the firm they eventually dumped the TR network and went 100mbit Ethernet everywhere.
Time4aNAP
Premium Member
join:2007-04-09
Des Plaines, IL

Time4aNAP to phoneboy2

Premium Member

to phoneboy2
said by phoneboy2 :

Token Ring was better in theory...
LOL... Reminds me of the die-hard Betamax or (pre-Intel) Macintosh zealots. They'd declare their product "better" by virtue of a single, arcane and marginal advantage, ignoring the competition's more practical advantages.

Seriously, if token ring hadn't been a product of Big Blue (who nobody got fired for buying), it would have been a non-starter.
jcf228
join:2001-09-05
Brooklyn, NY

jcf228

Member

Speaking of Token Ring

We STILL have a token ring network in the building where I work in downtown NYC (at a Fortune 100 Company). I have a Madge 4/16 card plugged into my laptop right now.

cbrigante2
Wait til Next Year
Premium Member
join:2002-11-22
North Aurora, IL

cbrigante2

Premium Member

Re: Speaking of Token Ring

said by jcf228:

We STILL have a token ring network in the building where I work in downtown NYC (at a Fortune 100 Company). I have a Madge 4/16 card plugged into my laptop right now.
I still have some sitting up above my server room.

alanhdsl
Premium Member
join:1999-10-09
Phoenix, AZ

alanhdsl

Premium Member

Re: Speaking of Token Ring

My office ditched token ring rather recently, so there's still some spare hardware around. Most of the older PCs still have the Madge cards. Once $300, they're now worthless. Madge, BTW, went out of business a couple of years ago.

PolarBear03
The bear formerly known as aaron8301
Premium Member
join:2005-01-03

PolarBear03 to jcf228

Premium Member

to jcf228
Pictures? Perhaps you could submit it to the Smithsonian?

cbrigante2
Wait til Next Year
Premium Member
join:2002-11-22
North Aurora, IL

cbrigante2

Premium Member

Re: Speaking of Token Ring

said by PolarBear03:

Pictures? Perhaps you could submit it to the Smithsonian?
Only if they hang it over Archie Bunker's chair!

phxmark
What Country Are We Living In?
join:2000-12-27
Glendale, AZ

phxmark to jcf228

Member

to jcf228
said by jcf228:

We STILL have a token ring network in the building where I work in downtown NYC (at a Fortune 100 Company). I have a Madge 4/16 card plugged into my laptop right now.
Ahhhh, such memories of Madge.

koitsu
MVM
join:2002-07-16
Mountain View, CA
Humax BGW320-500

1 edit

koitsu to jcf228

MVM

to jcf228
Token, token, who's got the token? (Only funny to those of us who know how frames get distributed on a TR net )

Shout-outs to my oldskool niggaz ARCNET and HPIB (still used apparently!).

--
Making life hard for others since 1977.
Time4aNAP
Premium Member
join:2007-04-09
Des Plaines, IL

Time4aNAP

Premium Member

Re: Speaking of Token Ring

said by koitsu:

Token, token, who's got the token? (Only funny to those of us who know how frames get distributed on a TR net )
Not so funny for those of us who were sent out to collect the lost tokens!
AMDonUT2004
join:2006-06-12
Bedford, VA

AMDonUT2004

Member

New network idea!

Sillystrings for networking, spray it everywhere, and have connectivity everywhere
Kord
join:2006-10-27

Kord

Member

Old School

I remember running the thinnet then some jerk took over and ran it with cheapo radio shack coax because he knew it would work. The next year we ran cat 5.

DaneJasper
Sonic.Net
Premium Member
join:2001-08-20
Santa Rosa, CA

DaneJasper

Premium Member

This is why we switch

When was the last time you used an Ethernet hub?

CSMA/CD is dead - and this memo points out it's limitations.

-Dane

•••
JimF
Premium Member
join:2003-06-15
Allentown, PA

JimF

Premium Member

Wrong is not necessarily stupid

I am surprised by the naiveté of those who think Mr. Bachrach was stupid just because he was wrong. Thomas Edison did not think that AC current would work either, and supported only DC. You could fill a book with such predictions. Small changes in economics or technology can change the picture entirely.
Time4aNAP
Premium Member
join:2007-04-09
Des Plaines, IL

Time4aNAP

Premium Member

Re: Wrong is not necessarily stupid

said by JimF:

...Thomas Edison did not think that AC current would work either, and supported only DC.
Edison knew very well that AC was superior for mass distribution. But his company was already committed to DC, with a huge financial investment in DC power plants. If it wasn't for that, and potential patent issues with Westinghouse if Edison did go AC, I would never have witnessed one of the last DC street light systems in my home town. (It had been converted to be fed from the AC grid by the time I was around.)

Given the business reality that Edison had to face, he did the best that he could, and used FUD marketing. The electric chair might never have been invented if not for this. They used to electrocute pigs to demonstrate how "unsafe" AC allegedly was. But connecting human death with Westinghouse's product had more impact.

Ironically, it was the success of Edison's light bulb that forced the Edison company to adopt AC...and a new marketing strategy.

Alakar
Facts do not cease to exist when ignored
join:2001-03-23
Milwaukee, WI

Alakar to JimF

Member

to JimF
said by JimF:

I am surprised by the naiveté of those who think Mr. Bachrach was stupid just because he was wrong. Thomas Edison did not think that AC current would work either, and supported only DC. You could fill a book with such predictions. Small changes in economics or technology can change the picture entirely.
A good current example; look at any thread on this sight discussing AT&T's Project Lightspeed.