I knew ATM was dead when the compromised on a 53 byte cell (I.e. frame) size.
Apart from the comical absurdity of the choice, the fact that they compromised was the canary that the standards process was a mess. Compare it to the IETF process that has been quite successful.
53 (48 payload + 5 header) was an efficient size for voice. There’s no fixed size that’s great for data — you always need fragmentation, so it doesn’t matter much what the exact cell size is. So although it seemed like an odd number, if you have to pick a fixed size it’s a good choice.
In most uses of ATM, a single VC (virtual circuit) carries all the data between a pair of switches. The protocol to set up a new VC is pretty expensive, so it's impractical to set up a VC for each TCP connection or even between each pair of endpoint hosts.
So mostly, VCs carry a stream of packets between switches, each with their own header.
ATM packets (cells?) are fixed size, but you can fragment higher level protocols across them. It's not common to run IP level fragmentation down to ATM sizes.
Good point but The battle continues with Google, Amazon, Facebook taking the corner of the Netherlands.
They run massive networks and definitely do it differently than the Bellheads
> ""How do you scare a Bellhead?" he begins. "First, show them something like RealAudio or IPhone. Then tell them that right now performance is bandwidth-limited, but that additional infrastructure is being deployed." Then "once they realize their precious voice is becoming just a simple application on data networks," point out "that the Internet is also witnessing explosive growth combined with 45 percent returns on investment, compared to 5 percent growth and only 12 percent or so returns for voice." In short," says Doran, make Bellheads realize they are witnessing their extinction."
That was definitely the real fight going on - control over calls and other services, with the ability to do differentiated quasi-monopoly pricing. Ultimately:
> "IP is hard to charge for? That's not a bug, that's a feature!"
This is the defining feature of the internet: you don't have to go through the billing system.
(If you want to see a telco-designed internet protocol that's in wide use, try SIP: the separation of control and data planes is absolutely characteristic.)
Real is still around. They occupy a floor or two of a small office building near the stadium in Seattle. From what I understand, they are essentially patent trolls now.
Few things ever cease to exist entirely. There's a hobbiest still running every flavor of VAX out there; still practicing practitioners of all sorts of weird cults. Real is no longer a software company nor can you realistically (no pun intended) use any of it's products. It has ceased to exist in all senses but historical, and the patent troll firm still bearing it's name should be seen as antiquated.
On the other hand, there's probably someone there who believes that they're still fighting the good fight against someone who cheated them, and they might be right. We've gotten so used to cheaters as a society that we really can't tell what's real achievement and what's fraud.
Kind of surprised by the number of words in here that I haven't heard in years, most of which I might add, signify mundane things, not specific technologies. Examples: "the Net," "netizen," "workstation"
As someone who was single-digit in years when this article was written, can anyone give a brief summary of how the fight turned out? Did the RISC-like sedimentation predicted in the article come to pass?
I had some ATM in Customer sites the late 90s as a LAN / backbone technology. By the early 2000s it was disappearing. IP over gigabit Ethernet and layer 3 switches with wire-speed routing replaced it everywhere I ever used it. ATM is still inside older DSL networks, but I haven't seen it used as a WAN technology in a long time.
I don't know enough about the innards of fast IP routers to say, but my gut says that, given the differences between IP and ATM, there's probably not much in the way of legacy ATM hiding inside our modern gear.
Insofar as the "Bellhead" vs. "Nethead" mentality, as another comment says, the majority of big "Nethead" ISPs have all been acquired by legacy telco interests (I wept inside when CenturyLink acquired Level 3, for example). In the service provider space, at least, I think that the "Bellheads" won.
There's a fossil of ATM surviving inside the ADSL protocol (although i suppose there aren't many people still using that) - ADSL has an ungodly stack of different protocols, but at some point, there's AAL5, an adaptation layer on top of ATM, which in turn sits above the physical layer. ATM cells are fixed-size and short, and AAL5 creates variable-sized long frames on top of them. PPP then uses that, and IP uses PPP.
Although from what i remember, the physical layer doesn't carry actual ATM cells; there's some scheme for packing them into large bundles, eliminating the repeated header, for more efficient transmission. So ATM is a sort of fiction which helps translate PPP messages into ADSL frames. It's been a long, long time since i really understood ADSL, though.
ATM cell size and phone data packet size relationship? Lots of people still think "more smaller is better" e.g. considering pMTU problems in IPv6, sending shorter non-fragmented IP packets is way better than sending long ones causing end-to-end congestion (no intermediate frag in V6)
Various things happened. Improvements made doing longest-prefix matching in hardware possible for IP routing. MPLS came along which allowed service separation and traffic engineering across a packet switched core.
Customers only wanted Ethernet/IP so the overhead of running an ATM layer in the middle didn’t make financial sense.
This was a seminal article of my youth. I still use the terms "Nethead" and "Bellhead" sometimes in conversation, and I refer people to the article fairly regularly. Setting aside the references to particular technologies, I think it gives a good background on a fundamental philosophical business difference.
The MCI hub in Colorado Springs in the 90s was like something out of a movie. And everyone that worked there seemed to think they were actors in a big movie. And the Northwestern Bell / US West / Qwest people were funny, too. Both sides really did talk and act like they were the rulers of the world, and actually hated each other with a passion, but I wonder how often they go to visit the graves of uu.net or others like it. Oh, that's right, most of them were laid off or quit decades ago.
As an old voice guy, it's still hilarious to me that AT&T and Verizon are the two biggest players in cell phones, which probably carry most of the data to end users. The IP guys didn't really figure that one out until it was too late. So they won the protocol fight, but didn't get paid.
> The IP guys didn't really figure that one out until it was too late. So they won the protocol fight, but didn't get paid.
AT&T collapsed and was reborn through mergers and slapping the facade name on a new entity, ending up in the present form as another soon to be bankrupt gigantic pile of debt with zero growth.
Meanwhile all the money went to the IP world: Google, Microsoft, Apple, Amazon, Facebook, Cisco, Qualcomm, Broadcom, Samsung, etc. The IP world saw all the value gain, AT&T and Verizon - once you tally the debt, debt interest, mergers - have seen zero inflation adjusted gain in either their businesses or their stock market values in 20 years.
> AT&T and Verizon are the two biggest players in cell phones
You're incorrectly assessing where most of the money is in phones, which led you to the incorrect end conclusion about AT&T and Verizon being the ones who got paid. Apple all by itself is deriving more value from cell phones than AT&T and Verizon combined. Google search via mobile is generating ~$30 billion per year in operating income, that's more than Verizon is making from its entire mobile segment. Facebook got nearly all of its ~$24 billion in operating income over the prior four quarters from mobile.
The IP/Netheads are the only ones who actually got paid.
They wouldn't be amused trying using that. Glacial speeds - so no multimedia; no search, no resources which became so traditional one takes them for granted, like maps, social media...
I actually enjoyed the internet more back then, or well, slightly after. I think I first went online about 1998. There was something magical about being a 9 year old fledgling programmer communicating via UDP, port scanning and connecting to services, and generally seeing what was out there.
I try to imagine it like space now. We're now looking out into space, seeing what is there. If in 1000 years we have it all mapped out and we just use Spoogle to locate something – "ore rich moon, ores:bauxite, distance:50LY" "1–243,123 results" – the fun of searching and discovering will be kind of lost.
Neat old story. Sean rocks. Bellheads lost except in lobbying. IP Everywhere is real. Screw the bellheads. These fools want you to put 5G uplinks in your house.
41 comments
[ 3.0 ms ] story [ 50.2 ms ] threadApart from the comical absurdity of the choice, the fact that they compromised was the canary that the standards process was a mess. Compare it to the IETF process that has been quite successful.
32 would have been a bad choice for data, as a large fraction of all internet data packets are 40-byte TCP ACKs.
So mostly, VCs carry a stream of packets between switches, each with their own header.
That was definitely the real fight going on - control over calls and other services, with the ability to do differentiated quasi-monopoly pricing. Ultimately:
> "IP is hard to charge for? That's not a bug, that's a feature!"
This is the defining feature of the internet: you don't have to go through the billing system.
(If you want to see a telco-designed internet protocol that's in wide use, try SIP: the separation of control and data planes is absolutely characteristic.)
I remember Real, but had forgotten the second - a VoIP phone "Not Cisco’s OR Apple’s" : http://www.appleiphonereview.com/issues/see-this-1996-iphone...
(first hit on search for "iphone 1996")
On the other hand, there's probably someone there who believes that they're still fighting the good fight against someone who cheated them, and they might be right. We've gotten so used to cheaters as a society that we really can't tell what's real achievement and what's fraud.
I had some ATM in Customer sites the late 90s as a LAN / backbone technology. By the early 2000s it was disappearing. IP over gigabit Ethernet and layer 3 switches with wire-speed routing replaced it everywhere I ever used it. ATM is still inside older DSL networks, but I haven't seen it used as a WAN technology in a long time.
I don't know enough about the innards of fast IP routers to say, but my gut says that, given the differences between IP and ATM, there's probably not much in the way of legacy ATM hiding inside our modern gear.
Insofar as the "Bellhead" vs. "Nethead" mentality, as another comment says, the majority of big "Nethead" ISPs have all been acquired by legacy telco interests (I wept inside when CenturyLink acquired Level 3, for example). In the service provider space, at least, I think that the "Bellheads" won.
Although from what i remember, the physical layer doesn't carry actual ATM cells; there's some scheme for packing them into large bundles, eliminating the repeated header, for more efficient transmission. So ATM is a sort of fiction which helps translate PPP messages into ADSL frames. It's been a long, long time since i really understood ADSL, though.
http://cs.hadassah.ac.il/staff/martin/Seminar/MPOA/mpoa.pdf
Customers only wanted Ethernet/IP so the overhead of running an ATM layer in the middle didn’t make financial sense.
As an old voice guy, it's still hilarious to me that AT&T and Verizon are the two biggest players in cell phones, which probably carry most of the data to end users. The IP guys didn't really figure that one out until it was too late. So they won the protocol fight, but didn't get paid.
AT&T collapsed and was reborn through mergers and slapping the facade name on a new entity, ending up in the present form as another soon to be bankrupt gigantic pile of debt with zero growth.
Meanwhile all the money went to the IP world: Google, Microsoft, Apple, Amazon, Facebook, Cisco, Qualcomm, Broadcom, Samsung, etc. The IP world saw all the value gain, AT&T and Verizon - once you tally the debt, debt interest, mergers - have seen zero inflation adjusted gain in either their businesses or their stock market values in 20 years.
> AT&T and Verizon are the two biggest players in cell phones
You're incorrectly assessing where most of the money is in phones, which led you to the incorrect end conclusion about AT&T and Verizon being the ones who got paid. Apple all by itself is deriving more value from cell phones than AT&T and Verizon combined. Google search via mobile is generating ~$30 billion per year in operating income, that's more than Verizon is making from its entire mobile segment. Facebook got nearly all of its ~$24 billion in operating income over the prior four quarters from mobile.
The IP/Netheads are the only ones who actually got paid.
I try to imagine it like space now. We're now looking out into space, seeing what is there. If in 1000 years we have it all mapped out and we just use Spoogle to locate something – "ore rich moon, ores:bauxite, distance:50LY" "1–243,123 results" – the fun of searching and discovering will be kind of lost.
[1] http://www.clock.org/~smd/new/
https://www.gofundme.com/seandoran