Network Geeks: How They Built the Internet

By Brian E Carpenter

The impact on modern society made by the Internet is immeasurable. Yet some questioned “why anyone would want such a thing” when the idea was first introduced.

Part history, part memoir and part cultural study, Network Geeks charts the creation of the Internet and the establishment of the Internet Engineering Task Force, from the viewpoint of a self-proclaimed geek who witnessed these developments first-hand. With boundless enthusiasm and abundant humour, Brian Carpenter leads the reader on a journey from post-war Britain to post-millennium New Zealand, describing how the Internet grew into today’s ubiquitous, global network, including the genesis of the World-Wide Web in the hotbeds of a particle collider at CERN. Illuminating the science and technology behind the apparent “magic trick” of the Internet, Network Geeks opens a window into the initially bewildering world of the Internet engineering geek. After reading this book, you may wish to join this world yourself.

• Language: English
• Publisher: Copernicus
• Release date: Apr 10, 2013
• ISBN: 9781447150251

Book preview

Brian E. CarpenterNetwork Geeks2013How They Built the Internet10.1007/978-1-4471-5025-1_1© Springer-Verlag London 2013

1. Hey Folks!

Brian E. Carpenter¹ 

(1)

Auckland, New Zealand

Abstract

Imagine that you are sitting in a very large room. It’s a hotel ballroom, with a grandiose name like Regency or Continental. It’s a square box with plain walls and no windows. The ceiling is high, three metres above you. There’s little decoration, the colours are neutral, and the double wooden doors at the back are only impressive when viewed from a distance. There’s an annoying background of air-conditioning noise, but the lighting isn’t harsh. There’s a generic, characterless carpet.

Imagine that you are sitting in a very large room. It’s a hotel ballroom, with a grandiose name like Regency or Continental. It’s a square box with plain walls and no windows. The ceiling is high, three metres above you. There’s little decoration, the colours are neutral, and the double wooden doors at the back are only impressive when viewed from a distance. There’s an annoying background of air-conditioning noise, but the lighting isn’t harsh. There’s a generic, characterless carpet.

You’re sitting on a plain stackable chair with a metal frame and a firm cloth-­covered seat. It’s hooked onto another chair at left and right. It’s one of a long row, part of a large block of chairs, and there are three or four such blocks separated by aisles. Altogether, about a thousand seats face a low stage at one end of the room. On stage, there’s a long table with 15 chairs behind it and a podium in the middle. There are microphones and water glasses on the table. Obviously, there’s going to be a panel of experts up there, and a lecturer at the podium.

At each side of the stage stands a large projector screen. It looks as if somebody’s holiday snapshots and cat pictures are being shown in sequence. A woman in casual clothes is messing around with a laptop computer on the podium; she must be a technician setting things up.

You start to look around you. In the aisles between the blocks of seating are microphone stands, so it seems as if audience participation is expected. But if this is a typical industry conference, where are the posters, banners and sponsorship logos to remind you where you are? As you wonder whether you’ve arrived an hour too soon, before the room has been properly set up, you are startled by what sounds like live music. Indeed it is; a few people have appeared in the front row with small drums and have started drumming together, in a pleasant and soothing rhythm. But why? What has this to do with the Internet? You thought you were attending a serious conference about Internet technical standards.

More people are drifting in. Here’s a busy-looking person in business clothes going up onto the stage. The woman already there looks up and says something to him. He presses a couple of buttons and says ‘One, two, three’ into a microphone; the loudspeakers are now on. This is all backwards – the casually dressed woman seems to be giving the orders, and the smart young man in a business suit is the hotel’s audio technician.

The people coming into the room are an odd looking bunch. Most of them are white males of various ages, but none of them are dressed for a serious meeting. Despite the air conditioning, several are wearing shorts and sandals. Many are wearing T-shirts, some of which appear to have a computer program printed on the back. One guy is wearing a tie-dye shirt that seems to be left over from the 1970s. There’s a scattering of grey or bald heads, many pairs of jeans, and a few men with long hair and beards that look older than the tie-dye. Is this a convention for superannuated hippies?

Ah, wait, here come a couple of men in proper business dress and ties. One of them is even wearing a neat grey three-piece suit. It’s funny, because he looks just like someone you once saw on the front cover of a magazine wearing a silly T-shirt. He’s talking animatedly with another man wearing a black suit and a Homburg hat, which he swiftly replaces with a yarmulke. Surely, these must at last be the people in charge. However, they wander down the aisle, greeting others as they go, and finally choose seats near the front.

A few of the people coming in are women, neatly but casually dressed. Quite a few of the men appear to be Asian, but there is only a scattering of black people. The buzz of conversation is in many languages. Even when it seems to be in English, the snatches you hear make little sense. ‘Yes, but what’s the threat model?’ … ‘You can’t announce a slash twenty-six and expect anyone to route it!’ … ‘I mean, he’s still trying to use MD5, what use is that?’

Almost everyone is carrying a laptop computer, and as they sit down, they scrabble around on the floor. You look down to understand why and see a line of interconnected blocks of power outlets snaking across the carpet. People are plugging in their laptops, and at least half of them are using adaptors for plugs from another country. As more and more people arrive, they construct small sculptures of adaptors plugged into adaptors and extension cords plugged into extension cords. By the time everyone is seated, the floor is a seething mass of cables and connectors. The conversation moves on: ‘What’s the SSID?’ … ‘Is IPv6-only working for you?’ … ‘Oh no, I lost my VPN again’. And most bizarrely ‘What’s the jabber room?’

The drumming from the front reaches a crescendo and stops, followed by scattered applause. It seems that the drummers were just regular participants in the meeting who like to drum. The cat photos on the big screens have been replaced by a slide saying ‘Welcome to Plenary’ and some letters and numbers. There is also a geeky-looking logo. The woman on stage is hovering near the speaker’s podium. The audience, if that’s the right way to describe such a ragtag assembly, is slowly settling. A tall blind man with long blond hair comes in, finds his way to the front row and sits down. He too has a laptop computer, on which he’s soon busy typing quickly, while talking to his neighbours, and apparently listening to the computer talking back to him through a small earpiece.

There’s a call for silence and some loud shushing. People still standing at the back find seats, and somebody closes the doors, which are immediately pushed open by latecomers, letting in bursts of loud talk and laughter from the lobby area. The woman at the front says a firm ‘Hello’ into the microphone, and the residual conversations subside. It finally dawns on you that she’s in charge.

Your mind drifts back to this morning. You arrived from the airport late last night, stumbled from the taxi to the front desk, checked in and went straight to bed, ignoring the raucous crowd in the hotel bar. This morning, you dressed smartly, as one does for a meeting, and went down on the dot of 8 o’clock to register and collect the promised breakfast. Registration was easy, but instead of a nice conference bag, all you got was a name badge, a Manilla envelope and a free T-shirt. Then you found yourself competing for breakfast items with several hundred other people, all going for the usual hotel bagels, breads and bananas. A few of them looked as lost as you felt, but most of them were deep in conversation as if they’d been there all night, or they were sitting on the floor deeply absorbed in their laptops, sipping coffee or orange juice and dropping crumbs freely onto their keyboards. Standing juggling your computer bag, your newly acquired Manilla envelope, your bagel plate and your coffee mug, you held a stilted conversation with someone from China who works for a very large company you had never heard of. As your bagel landed face down on the hotel carpet, he was telling you that he was here for sipping and simple, or at least that’s what you thought he said, and then he asked if you were going to boff. It made no sense whatever; in fact it seemed vaguely insulting, until you consulted the agenda in the Manilla envelope, to find out what SIPPING and SIMPLE and BOF meant.¹

Towards 9 o’clock the breakfast area began to clear out. You consulted the agenda again, to find that the session you planned to attend was in Mango 4, a featureless room that proved to contain about fifty seats but almost a hundred people. After a few minutes of confusion, somebody yelled out ‘We’re swapping with Plum 2’, and the hundred people all set off in search of the new room, meeting a smaller crowd coming the other way. After five minutes of pushing and shoving, the meeting finally started in the new room, with everyone seated this time. Someone wearing jeans and yet another silly-looking T-shirt said ‘OK, let’s get started, Jim’s doing the minutes, thanks Jim, but I need a jabber scribe…. Anybody?’ Mumbled conversations continued. ‘ANYBODY? JABBER?’ Finally somebody in the front row said ‘OK, I guess so’. Jeans man said ‘Thanks’. Eventually, you worked out that jabber is the name of an instant messaging system, and someone had just had their arm twisted to transcribe the session as instant messages for remote participants. Jeans man gave his name and then said ‘I’m chairing this session; the co-chair couldn’t make it, but she’s on jabber and will chip in when she needs to. Er, here’s the Note Well’. A page of legal-looking text flashed up on the screen, and as quickly vanished. ‘You’ve all seen it before. Usual rules for this meeting, please stick to the agenda, line up at the mike because if you speak from your seat the people on streaming can’t hear you’. You worked out that the whole session was being sent out as a live audio stream over the Internet. Whenever someone started speaking from their seat, there were yells of ‘Mike’ around the room, and the speaker lumbered across the room to the nearest microphone.

An agenda flashed up. But most people were staring down at their laptop screens; at least half of them seemed to be busy typing. People wandered in and out at the back. The jeans man ran through a tedious status report on several documents in progress. As he mentioned the third or fourth document, an overweight man in khaki shorts materialised at a microphone and said ‘Er, about the requirements draft, we said on the list that failover support wouldn’t be mandatory, but it’s still a MUST in draft seven’. Elsewhere in the room, someone said ‘Oh, right, sorry, I missed that. I’ll fix it in eight’.…‘Mike!’ Khaki shorts relayed that comment into the microphone. The jabber scribe at the front hurried to another microphone and said ‘Julia’s on jabber – she says that wasn’t a clear consensus on the list, we need to check’. Jeans man sighed ‘OK. Let’s test the opinion here: please raise your hand if you think failover support should be mandatory’. As a few hands goes up, a man in blue slacks rushes to the microphone and says ‘Wait just a moment. That would mean having at least two servers would be mandatory. That’s ridiculous for small deployments’. Jeans man sighed again. ‘OK, cancel that, please raise your hand if you think failover support should be mandatory if there’s more than one server’. Blue slacks retreated, with his hand up. Jeans man: ‘Now raise your hand if you think failover support should not be mandatory … now raise your hand if you don’t care … OK, looks like it isn’t mandatory, but we need to check on the list’. Blue slacks was back at the microphone: ‘So is it SHOULD or MAY?’ Jeans: ‘Sorry, you’re right … hands up for SHOULD … hands up for MAY … looks like a SHOULD. We’ll confirm that on the list before posting zero-eight’.

It took a while to decode that and understand what actually happened. The working group chair (jeans man) was mentioning recently revised documents to see if anyone in the room had objections. Khaki shorts was objecting that an agreed change had not been made. The editor of the document said he’d fix it in draft 8 of the document, but Julia was objecting via instant messaging that the change hadn’t obtained a clear consensus in the group. The chair started a show of hands to test the consensus, but blue slacks objected to the question. The chair refined the question, took a show of hands and announced the result, but stated that the consensus had to be tested on the working group’s email list, so that those who weren’t at the meeting could give their opinion too. Blue slacks objected again, because something that isn’t mandatory (MUST) could be either recommended (SHOULD) or merely optional (MAY). The chair took a show of hands on that point too and repeated that the consensus had to be tested on the email list.

Later in the day, the chair would send an email out to the working group’s mailing list, describing the questions asked at the meeting and the results of the shows of hands and asking for anyone who disagreed to reply to the list as soon as possible.

This was only the beginning of the meeting. Once the status review ended, there were a couple of presentations of new draft documents, with lively debates over abstruse technical details. Apart from a few diagrams, the presentations were very short – everyone participating in the debates had obviously read the documents closely and most of them were reading out loud the sentences they wanted to discuss from their own laptop screens. Quite a few comments were relayed to the microphone by the jabber scribe. Jeans man was constantly moderating the discussion, trying to focus the debate on yes/no choices whenever he could. At the end of both presentations, he asked ‘OK, with all those changes made, do we accept the draft as a working group item?’ These times, he asked for a ‘hum’ and meant it literally: people who wanted the group to officially adopt the draft hummed more or less loudly. When he asked ‘Anyone against?’ there was a loud, but solitary, hum against one of the drafts, from khaki shorts, who had given several articulate but incomprehensible arguments at the microphone. But both documents were adopted (Fig. 1.1).

A310448_1_En_1_Fig1_HTML.jpg

Fig. 1.1

A working group at the 85th IETF (Photo: B. E. Carpenter)

Your whole day, apart from a short lunch break as confusing as breakfast, was a sequence of such meetings. At every changeover, you had to choose between eight different parallel sessions, all described by obscure acronyms. Now, your brain is buzzing. Your mind seems to have turned to mush. And to think that most of the participants have been in similar meetings since Monday morning.

You resurface in the reality of the plenary session in the Regency Ballroom. You’ve missed the chairwoman’s introductory remarks, and now finally someone in proper business attire is speaking. He’s a vice-president from the meeting’s sponsor and is speaking about exciting market opportunities and how his company emphasises user satisfaction. Strangely, nobody in the room seems to be listening; most of them are concentrating on their laptop screens and keyboards, and there seems to be quite a bad attitude in the room. After ten minutes that seem like an hour, the VP flashes up a slightly technical diagram of his company’s products, mutters something about his people being available afterwards if there are any questions and stops speaking. There is scattered polite applause. The chairwoman thanks him wholeheartedly for the sponsorship – somehow she forgets to mention his speech – and hands over some sort of a plaque. The sponsorship money receives lively applause.

A baldish man hurries to the podium and rushes through some slides about the financing of the meeting, and some other administrative or legal-looking slides. At the end he says ‘Any questions?’ and makes to leave the stage as quickly as possible. During this, the laptop staring and bad attitude in the room has continued. But a white-haired fellow has appeared at one of the room microphones. (How do people manage to just appear at microphones, without apparent motion from elsewhere in the room?) He asks how much of the meeting revenue has been spent on cookies and coffee, but doesn’t seem satisfied with the answer. He’s obviously concerned about something much more fundamental than cookies, but what? It’s very puzzling.

Other people show up on stage and give short reports, mainly spoken in acronyms. There are one or two questions from the floor after each report. You notice that there seems to be a hard core of people who are paying attention to these administrative reports, and they can give as much acronym in their questions as the speakers do in their slides. These people, mainly men and mainly native speakers of English, are all in the oldest age group present and all seem to know each other. And they all seem concerned about something fundamental, whatever their actual question.

Finally the series of reports is over. The chairwoman says ‘OK, now for the fun part: Q and A. Will the Area Directors come up here, please? All of you, please?’ Fifteen or so people, mainly men, emerge from the front rows and take seats on the stage, behind the table and its microphones. They each state their name, followed by a word that appears to somehow relate to Internet technology, ‘transport’, ‘network’, ‘security’ and so on. Meanwhile, several people have lined up at the floor microphones. All over the room, heads and eyes are raised from the laptop computers. The chairwoman moderates a question and answer session. One or two of the questions expand on points from the various administrative reports, but she urges the room to move on to technical issues. The technical questions, when they come, are also framed almost entirely in acronyms, and the answers from the stage are the same. For example, there’s a long discussion between the audience and the people on stage about something called MD5. It’s repeatedly referred to as ‘broken’ and ‘unfixable’, but some people seem to say that this may be true, but it’s still much better than nothing.

As the clock moves on (and at this point you notice that the year is 1996), someone asks the following: ‘I have a question about scalability. We hear that traffic is doubling every 100 days. Is that true, and what does it mean for us here, because I don’t think any of our designs expected that?’ One of the people on stage, a stressed-­looking guy of above average weight, takes that question. ‘Well, at our network, we are seeing those growth rates right now. It’s scary, we don’t see how long it can continue, but with more and more people connecting and much more Ethernet and T1 stuff, not just dial-up, and cable modems coming along, it’s not going to stop just yet awhile. So we better plan on it, and that means, er, doubling capacity every three months for a while. Look, let’s just be happy about it, we’ve invented an industry. Hey folks, it’s like we’ve invented the telephone. But you’re fundamentally right, we have a scaling problem’.

Finally, long after the planned time, the discussion winds down. It’s after 10 p.m. Your brain is fried, but buzzing with new ideas. You’ve heard about more technical innovations today than in the whole of the past 12 months. You want nothing more than to crawl into bed. But you promised to meet some people in the hotel bar after the plenary meeting, to discuss an idea that a couple of you came up with during a coffee break in mid-afternoon. In the end, you will get to bed after three a.m., with more meetings starting at nine sharp tomorrow. By now you’ve agreed to write a draft about the new idea, and you’ve invented your very own acronym.

Congratulations! You’ve survived your first day at a meeting of the Internet Engineering Task Force, where the basic technical standards for the Internet come from.

This chapter described the imaginary experience of a newcomer to the Internet Engineering Task Force (IETF), at a meeting which is a fictional blend of many meetings. The Hey folks comment was actually made, more or less in the words suggested above, I think by Mike O’Dell from UUNET, at an IETF meeting in about 1996. It was unfortunate that the spectacular UUNET growth rate for a few months at that time somehow got transmuted into a myth about long-term growth of the whole Internet, until the bubble burst in 2000 and 2001. But Mike’s comment about the invention of an industry was right on the nail.

Footnotes

1

SIPPING and SIMPLE are the acronyms for two working groups, and BOF means a ‘birds of a feather’ meeting.

Brian E. CarpenterNetwork Geeks2013How They Built the Internet10.1007/978-1-4471-5025-1_2© Springer-Verlag London 2013

2. Before My Time

Brian E. Carpenter¹ 

(1)

Auckland, New Zealand

Abstract

Although calculating machines can be traced back as far as Archimedes and the Antikythera mechanism, and clever devices such as church clocks, barrel organs, and music boxes were widespread by the eighteenth century, machines that process information and machines that communicate with each other were both nineteenth-century inventions.

Although calculating machines can be traced back as far as Archimedes and the Antikythera mechanism, and clever devices such as church clocks, barrel organs, and music boxes were widespread by the eighteenth century, machines that process information and machines that communicate with each other were both nineteenth-­century inventions.

The first information processing machine was awarded a patent by Napoleon Bonaparte in 1804; it was the automatic loom invented by Joseph Marie Jacquard in Lyon. Today, we think of automated information processing being based exclusively on electronics, but Jacquard’s patent was issued exactly a century before J.A. Fleming invented the first electronic device, a thermionic diode, at University College London (UCL). Napoleon had inherited a conceit from the French royal family, namely, a fascination with silk cloth woven with multicoloured patterns. Jacquard’s genius was to realise that the painstaking and slow manual process of converting written weaving patterns into warp and weft could be encoded in the form of holes punched in rows and columns on heavy cards. These cards were linked together into a sort of chain, and fed through a device, attached to a loom, which detected the holes with mechanical feelers. These feelers were linked to hooks that raised and lowered warp threads of various colours, as the weft was shuttled backwards and forwards. Thus, the coloured warp threads laid down what we would today call a bitmap of pixels, on the raster defined by the weft. Indeed, the term of art in French for the raster on a computer screen is trame, exactly the word Jacquard used for the weft.

Charles Babbage, usually credited as the first great inventor of the computer age, was well aware of the importance of Jacquard’s invention. He even owned a woven version of a portrait of the Frenchman that had been laboriously digitised in the form of Jacquard cards and then quickly woven on a suitably equipped loom. This was, arguably, the first example of a bitmap image produced by a machine.

Jacquard was a pragmatist. He fought on the losing side of the French revolution – Lyon, as a major silk-producing centre, tended to support the ancien régime which bought most of their product. He managed to sneak quietly back home when he proved to have backed the wrong horse. Babbage, like his twentieth-century successor Alan Turing, was more of an idealist, with less grasp of the practicalities of getting an invention funded and up and running. Thus, apart from its wild success in the weaving industry, the Jacquard card had to wait for another pragmatist, Herman Hollerith, to become clearly identified as a