Nothing is what it seems

The sheer speed of transactions available through telematics, the coupling of computers with telecommunications, notes political studies lecturer Joe Atkinson, has transformed business worldwide, virtually obliterating most local markets and making the international financial markets much more volatile. In all facets of communications, computers will be big. Their components on the other hand will become even smaller… With hardware getting cheaper it won’t be long before there’s a personal computer in most homes. Soon, said Stella Belliss of the DSIR’s information technology division, we’ll be able to tap into all sorts of computer networks and database systems. Which means we’ll be able not only to work and play Space Invaders but also do the banking and order the groceries on the same machine. Metro Magazine, 2020 Vision, August 1990 by Peter Allison.

From 1984, the Labour Government under David Lange began to introduce “the most sweeping reforms the New Zealand economy had seen in 50 years,”[1] effectively experimenting so much with political and business structures that internationally the country began to be viewed as a social laboratory.

The rapid reforms and their wide-ranging impacts included redefining the role of government and requiring the public sector to become more effective, transparent, and accountable. With these measures, it went further, faster than any other nation in the free world.[2]

In a few short years the culture went from one of regular government intervention, assistance, and subsidies to ensure level playing fields, to letting the market decide. The rationale was that the economy badly needed a shot in the arm. The adrenaline the social engineers prescribed was tough love, a kind of economic Darwinism where the realities of the marketplace would decide who succeeded and who failed. From 1986 that meant wide-ranging privatisation of state assets and a new aggressive, hands-off environment where railways, health, electricity, broadcasting, telecommunications, and even universities and research and development had to be profitable in the free market.

The SOEs Bill was introduced to parliament on 30 September 1986 and passed into law in a little over two months. It established nine new SOEs: Airways, Coal, Electricity, Forestry and Land corporations, Government Property Services, NZ Post and Telecom.

The main thrust of the reforms was economic restructuring through deregulation, corporatisation, and privatisation. The currency was floated and banks deregulated. All farming subsidies were removed. In the public sector, the SOEs Act identified the trading activities of government entities and separated them out, as being the role of businesses. The ‘rump’ government departments were then reorganised to become more efficient by means of the State Sector Act, which gave public servants more power to manage but demanded more accountability. SOEs were designed to be market driven and corporatised to introduce business disciplines and financial accountability. Corporatisation was the step before privatisation. The prevailing view was that public services by definition could never be commercially competitive; therefore governments should adopt a minimalist approach. The first consideration under the new business regime for SOEs was to prune staff through massive redundancies.[3]

The NZPO was split into state trading companies including NZ Post, Postbank and Telecom Corporation. On 31 March 1987, the newly created Telecom became an SOE. The government-owned Telecom Corporation then purchased the telecommunications-related assets of the old Post Office for $3.2 billion and underwent a major shift, from a public service organisation managing an essential public infrastructure to a commercially focused business.

There were still 38,000 party line customers on the Telecom network but work began in earnest to improve the network and the services offered, as the telecommunications market was progressively deregulated. Telecom launched its 025 mobile voice and CDPD analogue mobile data network, and by the end of 1987 it had 2000 customers. It cost $2.84 for a three-minute fixed line toll call between Auckland and Wellington and $4.20 for a three-minute off-peak call to Australia.

While the new Telecom was busy with internal restructuring and gearing up to impress potential investors, Murray Milner, who had been with the government department from 1971, was appointed as head of the Advanced Technology Group. Milner had joined the Post Office straight from school and done a bachelor’s degree and PhD in electrical engineering at Canterbury University. He was recognised early on as an important asset. When the news came through that he’d been appointed to head an elite group of technologists to help Telecom prepare for the technical transitions ahead, he was in Silicon Valley, in California.

After spending a year at Carnegie Melon University he was part-way through a Harkness Fellowship at Stanford, completing postdoctoral studies in electrical engineering and public policy economics, based around his involvement in domestic satellite communications in New Zealand.[4]

From 1987 Milner would log on to the fledgling Internet at 2.4kbit/sec from Silicon Valley using his dual floppy disk laptop to keep in touch with the rest of the Telecom Advanced Technology Group. His mail would come into Waikato University, then on to Victoria University. His team would dial in for one hour a day to upload any messages or files. “We hired a lot of people by placing postings on technical newsgroups and while we were very vocal about where this technology might lead, within Telecom there were always issues raised about the underlying technology and the usability. There was little interest beyond the academic community,” Milner admits.

From the early days of the mainframe computer some form of messaging from the central machine out to connected terminals was possible, but it took years to break the proprietary hold so networks could connect to networks. Once New Zealand universities figured out how to link with international servers, to access academic and technical documents and download free code, there was an unstoppable momentum. While the backroom boys at Telecom were well aware of the technology and continued to use it, nobody quite knew for sure where TCP/IP would end up. It was nearly a decade before Telecom took the Internet seriously as a commercial opportunity, although it kept a close eye on the universities as they experimented with offshore connections. There was a growing concern about all this anarchy that seemed to bypass the established telephone networks.

Can anyone plug us in?

John Houlker from Waikato University and Professor John Hine from Victoria University’s Computer Science Department received an important invitation to attend an academic workshop in Princeton in late 1987. The invitation from Larry LandWeber,[5] who ran the International Liaison programme for CSnet, was the opening for a landmark moment in the history of the Internet in New Zealand. Waikato had already connected to the Internet backbone via CSnet and was technically operating the country code. But having been convinced by Telecom the way forward was X.25, it was still trying to cobble together an efficient solution to share that store-and-forward access with other universities. At the workshop Houlker was working the room, looking for a way to get software that would enable TCP/IP to run more efficiently over X.25, or perhaps convince someone to subsidise a circuit to the United States so that New Zealand could connect directly to the Internet backbone.

He wasn’t having much luck. Then at a workshop dinner he was approached by Tony Villasenor,[6] head of National Aeronautics and Space Administration’s (NASA) scientific Internet, and Jim Hart, a high-performance computing specialist from NASA’s Ames Research Centre. They had noticed Houlker’s persistent campaigning and thought they might have common interests. “I thought we’d struck gold. They had been working on a programmer at NASA to get the Internet across to the Asia-Pacific region to connect a variety of science projects they were involved in. They typically used leased circuits on a project by project basis but finding someone in the other country to help them sort out the options, and the cost of getting all that equipment in and running was time consuming and expensive,” said Houlker.

Villasenor and Hart were heavily involved in ARPANET and linked directly into the NSF Internet backbone, and were looking at ways to avoid the cost and complication of employing people in each country. They had several projects in New Zealand, including the Kiuper Airborne Observatory in Christchurch, and a weather station at Lauder in Central Otago, as part of an Earth Resources 2 (ER2) project which involved U2 spy planes, gathering ozone readings over Antarctica. The Kuiper base tracked a converted US Starlifter aircraft with an infrared telescope in the fuselage, which circled the Tasman at high altitude, plotting certain stars to provide critical data. The U2 aircraft were based at Christchurch Airport along with their original pilots, but required highly accurate and classified meteorological data from the United Kingdom, relayed through the New Zealand NASA link. A key location for all NASA’s activity in New Zealand was the NSF Antarctica Gateway in Christchurch.

Villasenor and Hart were not only helping out the NSF with its projects in New Zealand but were hoping to use this association with New Zealand to convince the science and academic community in Japan to join the Internet. “In fact NASA was very keen to convince Japan to go the TCP/IP way rather than OSI which they were considering at the time. They thought that if New Zealand, Australia and other Asia Pacific connected to NASA’s Pacific Communications programme (PACCOM) using TCP/IP this might encourage Japan to head in the same direction.”

NASA agreed to pay for half of the circuit to New Zealand as part of PACCOM. Managing the remote end would be Dr Torben Nielsen,[7] director of the Computer Science Department at the University of Hawaii. This was a marriage made in heaven because the University of Hawaii had the job of outreaching to the Pacific and endeavouring to increase the speed of its link to the NSF backbone. Nielson was championing that; also, because the ANZCAN undersea cable went through Hawaii, it made sense. It looked like a done deal, until representatives of the Australian research and academic networks weighed in with their own proposition.

They suggested New Zealand was still disorganised and Australia could deliver everything NASA wanted. If the link terminated in Australia they would sort out New Zealand from their end. Seeing the groundbreaking deal fading before his eyes, Houlker countered that the cost of a circuit between Australia and New Zealand was about the same as from New Zealand to Hawaii. “It may look a small gap between the two countries on the map but getting a circuit across the Tasman is not a trivial exercise. We were also able to demonstrate that the DSIR, the universities and the technical arm of Ministry of Agriculture and Fisheries (MAFtech) did in fact have a common view.”

Having to connect through Australia would not have gone down well. “Getting funding assistance from the US Government through NASA was very tantalising, and being forced to go through Australia and paying the full rate did not appeal,” Houlker said. NASA needed someone to take all the risk and at the New Zealand end it took some months to sort out the logistics. Mark Topping, computer centre director at the University of Waikato, called a meeting of directors from other universities where Houlker made a presentation, explaining his coup in securing a 9.6kbit/sec link to the Internet backbone.

It was going to cost $120,000 a year, or, with the NASA subsidy, $10,000 for each of the six main universities. Topping convinced Waikato’s vice chancellor to be guarantor and collect the money for the link into Waikato. In the end the universities, the DSIR, and MAFtec agreed to help cover the cost of the circuit to Hawaii. “It doesn’t seem much now but at the time that was a huge hurdle and the cost was a little more than we had been paying in total for international email in the different forms we got it,” said Houlker.

The initial management responsibility for the new international link was based around the university network group representing Spearnet (South Pacific Education and Research Network), which had been formed to manage access to Coloured Book services like Janet in the United Kingdom. Rather than the previous arrangement with CSnet, which ran the DNS on Waikato’s behalf, Waikato University would now run the DNS itself. Houlker was registered as the person responsible for administering the zone, and was entered in the WHOIS database as the contact for the country code, representing the University of Waikato. “While personal connections and trust were certainly part of the FOJ (Friends of Jon [Postel]) network of confidence, I would have expected a successor at the university to have taken my place if I had vacated the role.”

It’s alive… well, almost

Milo Madin[8], head networking engineer at NASA, had a penchant for using military terms for Internet use, and is credited with pioneering the use of the term ‘demilitarised zone’ for Internet interconnection. He had also used the term Forward Edge of the Battle Area (FEBA) in 1988 when naming the first Internet exchanges in the world; FEBA East at the University of Maryland, and FEBA West at NASA’s Ames Research Centre.

A couple of years later, senior management at the agencies became aware of what the exchanges were doing, and while impressed at what was being achieved, were alarmed at the names. This was immediately resolved by re-designating them as Federal Internet Exchanges or FIX East and FIX West. According to Houlker exchanges were created because of “the complex and random interconnection required between the various federal networks and the NSFnet backbone and the ongoing tensions and differences of opinions between the agencies and the IT people managing the networks.” He based the original exchange at the University of Waikato on the FIX model.

In April 1989, a full 18 months after the Princeton meeting, the analogue undersea link to Waikato went live. At the Waikato end was a software-based gateway linking the various types of email, and a translation gateway built from an innovative mix of the CSnet software and code from the United Kingdom. Everything seemed to be working fine but when they switched on the 9.6kbit/sec analogue modem and the Proteon router provided by NASA, not a lot happened. The router had been misconfigured so Houlker soon had the top off and was poking around. Madin had sent the router without a manual because he was determined Houlker shouldn’t open the box or touch anything. Houlker eventually got Madin on the phone reading out instructions from the manual to get a clearer picture on how to move a chip inside the Proteon. “I remember it was a Sunday and I was a bit worried because there was no support in New Zealand, and if I broke it we’d have to send it back to the US for repair and still be paying all this money for an unused circuit.”

He carefully lifted the chip out of the router and on attempting to replace it, bent a pin. Thinking he’d blown the job, he tried one last option, gently straightening the pin, all the time fearing the worst, and thinking about the six-week delay and the huge bill if it broke. Fortunately the IC chip slotted securely into its new location. The router was reset and immediately there was action on the line. At first it was pings[9] back and forward to determine the remote location and the speed of the network, and packet tracing to see what was out there. The next phase used Telnet[10] to connect to the computer in Hawaii and see what addresses came up. One of the first messages that arrived on his screen was NORAD, which had Houlker presuming he had hit the North American Aerospace Defence Command headquarters. He’d only been on the US backbone a couple of hours and was now concerned he might be in trouble for trying to get into some US military site.

He soon discovered this was the handiwork of NASA engineer Milo Madin, who had called the humble old name server computer at the University of Hawaii NORAD. Pretty soon, email, the application everyone was eagerly awaiting, was up and running.

Part of the deal with NASA was that Waikato, and Houlker in particular, would be required to be the troubleshooter if things went wrong at the NSF operations in Christchurch, particularly if the connection wasn’t up to par. He often found himself on call, sorting out problems using remote technology, or flying to Christchurch or meet with the local technical people who didn’t have much networking experience. If something did go wrong Houlker was required to report back to the United States in detail. All costs involved were borne by Waikato University.

“One of the fibre cables across the Atlantic got broken and the connection was relayed onto satellite but the performance wasn’t good enough so I had to sort that out. Then they had kit that had been sent out from NSF in Washington which was designed to do back-ups of their Sun workstations over a local area network. They were wondering why the network was performing dreadfully and I eventually tracked it down to the fact that the workstations were trying to do their daily back-ups to Washington DC,” said Houlker.

Delays continue

Access to this important international communications resource was soon fuelling breakthrough collaborations in the United States and the United Kingdom, and the technical pioneers at New Zealand universities were again looking at ways to connect their network resources and create a national infrastructure. While Waikato was more organised in the way it dealt with the international link, the universities were struggling to get their slice of raw Internet. The absence of any government or other leadership from the 1970s onwards meant first attempts were ad hoc.

Frank March, who had moved to Victoria University as director of Computer Services after a decade at the DSIR and a spell consulting within government, said the lack of leadership was a major obstacle. “Nobody in authority and nobody at the general manager or even the normal managerial level had any idea what this technology was going to be able to achieve. It was driven outside the normal budget processes by individual enthusiasm, based very much on the fact that there were close linkages between computer science departments who knew about this technology, were working on it and reading the technical papers.”

This resulted in decisions being made about domain names and other technical and political issues that in hindsight might have been made differently. “They were trying to make important but arbitrary decisions with limited information because something was going to happen tomorrow that required a decision today. And those critical decisions were made by a handful of people.”

Auckland University had been in discussions with a Canadian university, which agreed to hook it up to Bitnet. After getting the connection to work using Bitnet software, emulating Bluebook Ethernet over Pacnet and running RSCV, it was ready to go live. Just as the university was in the final stages of sorting out the political issues, news got out that Waikato University had achieved what everyone thought was impossible: a direct connection to the US Internet backbone.[11] Auckland dropped everything in the hope of seeing a university-wide network emerge but true to form, nothing was easy or fast. Frustration was mounting as attempts were made to interconnect a confusion of networks and protocols to comply with the more open sourced TCP/IP of the wider Internet.

The first IP link from Waikato was a serial line interface protocol (SLIP) link between Victoria and Waikato running over the DSIR’s internal serial network, with an 8086-based PC-router at each end. It ran at 9.6kbit/sec, although the DSIR constrained it to only 4.8kbit/sec.[12] The theory was that there was sufficient spare capacity on the DSIR 4.8kbit/sec network to accommodate DSIR internal traffic and the university use. This proved not to be the case.[13] The link between Victoria and Massey University finally went in six months after the international link was established at Waikato, after the Wellington end got fed up waiting for the computer centres to get something working.[14]

While they continued to pay their share, it was still some time before Canterbury, Massey, Lincoln, and Otago got their promised link to the US backbone via Waikato. Efforts to try and tunnel TCP/IP traffic through the DSIR network had failed. “Initially we had a gateway to the Coloured Book software but this wasn’t as powerful as a direct connection to the entire New Zealand network. I was working on modifying TCP/IP software from Carnegie Mellon so we could use X.25 the same way we had on the CSnet. I had connected a trial network across the DSIR network between Waikato and Victoria but had huge trouble getting it to work,” said Houlker.

He persisted with the DSIR engineers, asking if there was anything wrong with their software. They reassured him it was rock solid. “I kept going back to what I was doing but couldn’t find much wrong. In the end I thought I’d try DECnet over the DSIR network and that didn’t work either but it did give me a pile of diagnostics showing faults with the DSIR’s X.25 system. By then the other universities were getting really annoyed with me over the time it was taking to get the network sorted. Then DSIR conceded that perhaps their X.25 didn’t work properly after all.”

Frank March admits there were enormous battles going on, that even today have left scars across the research and academic community. “It was so expensive to run lines up and down the country under a strong Telecom monopoly, so the idea was to use the DSIR network to connect the universities together by tunnelling TCP/IP through the DSIR NodeCode protocols. I don’t think John Houlker quite understood how the DSIR network was supposed to handle tunnelling, but the real problem was the DSIR network was appallingly slow and there was bad blood generated on both sides because of this.”

While some were keen or co-operative, others remained deeply suspicious at the universities plans. This was partly cultural and partly because the DSIR network failed to bridge the gap. “On paper it looked all right but it never did work properly. I recall years later talking to someone from DSIR who was railing against Waikato University and everything it stood for. This guy was still bitterly angry at what had happened. And people like John Houlker still shake their heads in disappointment. I suspect there was credit in what both parties were trying to do but it was just too hard,” said March.

All the while the pressure was on for academics to be able to email their colleagues around the world and to access the news. Bulletin boards had started to crop up from the late 1980s and were creating a lot of interest, making newsgroup content available to their users. Anything that could be cobbled together to make this work was acceptable at the time, and the DSIR and universities were a big part of that. Eventually the networks came together but long-term decisions still needed to be made about the protocols that would be adopted for the research and academic network and for the future. “That’s where the real difficulties came in, with people saying ‘my way is better than yours,’” recalled March.

The new, more commercial approach of state-owned Telecom resulted in a change not only in technology direction but in its attitude towards its clients, including the nation’s research and development facilities and universities. Instead of working with them for the kind of public-good goals expected of government departments, Telecom began playing a game of divide and conquer. It was clearly now placing profit ahead of the longer-term goal of a robust, nationwide academic and research network.

While the X.25 tunnelling problem was being resolved it suddenly became clear that a lot of time and effort had been wasted because Telecom did a complete about-turn in its charging. Its promises to the universities that X.25 would be delivered at a much lower cost than leased lines didn’t pan out.

Houlker said Telecom’s decisions on how it priced its different services was crucial. “They pushed us in one direction and with deregulation there was a 180-degree change to align with what was going on in the United States. Public X.25 charges stayed the same and they halved the cost of leased lines, which continued to drop. In fact the whole equation of trying to do stuff on X.25 dropped away so we planned on building our own leased line network.”

Within six months of the Waikato gateway going live, pure economics forced Waikato to review its connection with Hawaii. The network had become congested but increasing the bandwidth required additional multiplexed circuits and Telecom’s price for this was onerous. There was no advantage in remaining connected through the ANZCAN cable to Hawaii. Despite the inherent delays, the cost of a satellite circuit direct to the NASA’s Ames Research Centre in Mountain View, California, was slightly more affordable and gave better performance.

NASA’s Hawaii-based science Internet project was an extremely well-connected part of the US Internet with a direct link into NSFnet so any improvement in connection in New Zealand was going to be beneficial. Milo Madin, who managed the Hawaii gateway, sent an upgraded 14.4kbit/sec satellite modem, valued at $NZ7000. That transition didn’t happen as smoothly as Houlker had hoped either. Telecom insisted it must be ‘type approved’ and power tested at a cost of $3000. He had a few words to contacts at DSIR, which did type approving for Telecom, and the bill came back at $300. In October 1989, not long after the new modem went live, the San Francisco earthquake hit. Within four hours, when the batteries at the San Francisco telephone exchange ran out, the line went dead.[15]

Welcome to network limbo

Irishman Andy Linton was a late starter in computer science, having worked as an outdoor pursuits instructor and maths teacher in schools and the British Royal Navy. At the age of 30 he returned to the University of Newcastle to gain his master’s degree. In 1981, when computer science was still in its early days, Linton caught the bug and stayed on as a research associate on a project called Highly Reliable Distributed Systems, writing code to network Unix systems. His project was geared around building systems for air traffic control and train signalling systems. “We had multiple machines talking to each other and running complex algorithms; this was triple modular redundancy stuff which went into production systems eventually.”

His team knew of the TCP communications protocol but preferred to write their own drivers and build systems from scratch using Unix seventh edition. In 1982 he got hold of the first copies of the Berkley 4.2 BSD Unix code, which included TCP/IP. However the UK Government and various standards bodies discouraged its use. “They even had a joint network team, affectionately known as the protocol police; universities and other government departments looking to network computers were told to base all their development on ‘the purer’ seven layer OSI model,” said Linton. He and his associates were in a bizarre situation where all the interesting research they were doing was based on TCP/IP but this had to remain unofficial or they wouldn’t have been eligible for grant money. Once the unofficial TCP/IP stack was up and running across the research departments, Linton became the postmaster and applied for all the IP addresses for the university.

As part of his work in the computer lab at the University of Newcastle, Linton ended up supervising some of the masters students. Among them was a New Zealander, Cindy Treloar, studying computing software and systems design. She was heading back to New Zealand when Linton quipped, “If you hear of any interesting jobs let me know.” He’d been looking to advance his computer science interests and been offered a position with the University of Queensland the year before. Then his father passed away and he decided not to go. Cindy knew someone at VUW, who asked Linton to mail his CV. That resulted in a ‘bizarre phone call,’ more or less asking, “When can you start?” “I’m saying, ‘Well, you know if I resign now I could be there by August.’ And my wife Sheila’s sitting on the other side of the living room with a surprised look on her face, saying ‘What are you talking about?’”

There was a strong link between Newcastle and Victoria University, which Linton didn’t know about at that stage. American John Hine, Victoria’s head of Computer Science, had worked at Newcastle University before arriving in New Zealand. So, in 1989, with the university agreeing to pay the moving costs, Linton committed himself for three years. One of the benefits offered to him was full Internet connectivity. On arriving in Wellington, Linton, having enjoyed liberal email communication with colleagues across the United Kingdom and United States, learned full connectivity to the Internet was still being worked on. “They had explored a number of UK-style conventions and knew TCP/IP was the way forward but there was no plan on how to get there,” he said. Computer Science Department technician Mark Davies gave him the grand tour and Linton asked about the ‘full Internet access’ that had been promised. The response came back: “Well, you know… soon.”

Once access was cobbled together the Victoria Computer Science team couldn’t wait to download anything and everything. “We connected to the FTP server on US-based UUNET26 and began downloading files, including every bit of code we could get our hands on that would be useful to our various projects.” Victoria’s UUCP connections also enabled it to dial into the University of Melbourne to access the machine of Robert Elz,[16] an Australian Internet pioneer, who provided feeds to all sorts of interesting groups. “We burnt the link hot for a while.”

Grand ideas had been floated about how all the universities should be connected but the mix of different computers, protocols, and networking approaches, and the cultural clashes between DSIR and the universities, were keeping everyone at arms length. If you wanted to get access to the main universities, it was a two-stage process. To pull files into the country or from another university you had to bring them into Waikato University, which by then had a direct connection to the NSFnet Internet backbone. Then you had to use something different to copy files across and bring them down to Victoria using DECnet over X.25.

Distracted by their newfound connectivity and the ability to download as much open source code as possible, the enthusiastic pioneers at Victoria University’s Computer Science Department were about to face fiscal reality. “It seemed we had, ah, run up a large phone bill and basically got the message that we’d better knock our activities on the head. We were about $10,000 in the hole just through the network traffic,” recalled Linton. While Computer Sciences Department head John Hine was ‘very supportive’ and keen to see international connectivity, there was still the matter of who was going to pay the bill. No one in the accounting department was ready to even consider it could be written off as justifiable research costs. “The whole idea that you’d spend a lot of money on communications was relatively hard for people to grasp,” said Linton.

One way to avoid the high cost of connecting through UUCP in the United States was to ramp up efforts for Victoria to get an IP connection to Waikato University through the DSIR and charge for services to third parties. “Once you got to Waikato you could start piggybacking on NSFnet. That was a strong driver.” At this stage, even though the universities were all contributing to the cost of the new international link, the technical reality was still being worked through. There had been talk about running TCP/IP over DECnet over X.25 but it all seemed very convoluted and complicated. So between Linton, technician Mark Davies, and Jonathan Stone,[17] a solution was cobbled together to create direct channel circuits through the DSIR network using PCroute,[18] which acted as a cheap TCP/IP router. “There were some interesting challenges because the DSIR network was designed for terminal emulation and there needed to be some changes. Every time CTRL+P appeared in the data stream it took you out to a menu, and the whole system would lock up,” said Linton.

The trio then used their influence with various computer science departments to get a more direct link from Victoria University to Waikato, hopping from the DSIR network node at Victoria campus through to Gracefield Laboratories in Wellington where the DSIR had its IT support department. They had established the initial DSIR-based links to the international gateway at Waikato, using a modified form of SLIP over the DSIR circuits. Tony van der Peet[19] the network manager at DSIR, made the required changes to the DSIR’s Ace[20] boxes to allow the permanent links to be established, initially at the glorious speed of 48kbit/sec. “I think that was a real catalyst for the other universities. Others may have a different view but I believe that was the wake-up call,” said Linton.

Up until then the administration for handling the charging for the international gateway and attempting to co-ordinate a national research and science network had been in the hands of the old Spearnet committee. Now, with links cobbled together across the country and some form of network in place, a meeting was called between the seven university computer centre directors. Seats were reserved for Victoria, one each for the Computer Sciences Department and Computer Services. “I think that said something about us having established our credentials and that people wanted to listen to our ideas.”

It was still mostly computer scientists engaged in research and development who were interested in the Internet, although those with other science backgrounds were beginning to take notice of the opportunities it offered. The main one was access to the Usenet newsgroups and the discussion was centred around the massive volumes of data being downloaded – how this could be managed with limited disk space, and how the costs could be shared?

Auckland, Lincoln, and Otago still hadn’t connected to the Waikato gateway, but the meeting showed them what they were missing out on. “At this stage we were saying to Telecom that this [Internet] stuff was going to come really quickly and bite them in the arse if they didn’t get up to speed, but they were of the opinion that it would never catch on.” The universities were looking for a technology that could connect them all over a common set of protocols but Telecom insisted that because the Internet wasn’t X.25 compliant, it wouldn’t be reliable. “They were defending their proprietary networking approaches and saying no one would buy this Internet thing because of the unreliability of TCP/IP packets. They just weren’t on to it; their head wasn’t in that space at all,” recalled Linton. Rather than waiting for Telecom to catch up with the global trend, the universities were again forced into do-it-yourself mode.

Australia had connected to the PACCOM network a month after New Zealand, offering its users Internet access at flat-rate charges. Across the Tasman when the link got overloaded this was simply used as leverage to acquire further government funding for more bandwidth. Meanwhile Waikato had to come up with a strategy for equitable cost sharing of the international link with those now accessing the services on the university backbone. With no government funding to help carry the cost, the New Zealand administrative committee, after much discussion, agreed on a user pays approach.

John Houlker knew costs would only increase as demand grew, so early on a meter for volume billing was devised. He had read various white papers that claimed there was no way to bill by packet usage. “I couldn’t see any reason why it wouldn’t work so I drafted the specifications and my team produced the code fairly quickly.” The development proved hugely beneficial to Waikato in managing growth and charging users, but to the rest of the world it became a source of controversy. “This wasn’t the Internet way, “ was the feedback from those who had the benefit of strategic government planning and funding. Then, according to Houlker, when Chile used New Zealand as a model for its packet charging, academic papers began doing the rounds saying “the poison from New Zealand is spreading.”

Joining the dots

Victoria University academics continued to email like-minded colleagues around the world through the gateways being maintained by the computer scientists, who were determined to drive this networking forward, regardless of how slowly their bureaucracies worked. While struggling to sort out decent access to the Waikato gateway, Victoria had its own internal networking dilemma. The Computer Science Department was forced to move into its new campus building away from its computers. Mike Newbery and John Hine submitted a report recommending Ethernet routing rather than bridging technology to link the new Computer Science building to the existing Computer Centre. “You could tell from the report and looking at what was happening in the industry that this was the way things were going to go. It was a pretty safe bet although we had to fight against some strong opposition,” said former computer centre director, Frank March.

Victoria University eventually agreed to purchase routers from an unknown company called Cisco, rather than the better known Wellfleet or Proteon boxes, and to have them imported and supported by ECL (a division of the TV repair company Tisco),[21] from whom they were purchasing TRW access control or terminal servers. Victoria was the first in the country to buy Cisco equipment; it ordered five Cisco MGS routers, but Cisco was unable to deliver them on time and substituted the larger AGS models for three of them, enabling the creation of a much more robust and capable campus-wide network than planned.

John Houlker at the Waikato international gateway was also looking at this new company and the flexibility of its routers. “I spent quite a lot of time arguing that we should use Cisco routers instead of Proteon because they had other features we wanted, but I couldn’t get the folks at NASA to agree. After a while I decided Cisco was technically just as good as the Proteon, and rather than argue we left the gateway as Proteon but used Cisco routers for the network we built for the universities. Cisco routers at the end of Telecom’s point-to-point leased lines began rapidly appearing around the country as foundation stones for the first nationwide IP-based academic and research network.”

Grass-roots networking

The Ministry of Agriculture and Fisheries (MAF) had established its own MAFnet X.25 network in 1985. It had been quite advanced in its approach and an early adopter of WAN, initially based on Prime computers. Then in the 1980s its Invermay campus purchased the first DEC VAX computer, which saw MAF going entirely with Digital. It used X.25 for international communications through a Telecom gateway and X.29 for remote access.

Demand for linking with international research groups was growing but X.25 was expensive. As international collaboration became increasingly essential to keep up with international research, MAFnet found it was having to deal with X.25 OSI protocols for European connections, and as collaboration with the United States grew there was a need to enter the IP world. The difficulty of having to operate in both protocols resulted in a growing association with Waikato University and its international gateway, first initiated by its Ruakura Research Centre in Hamilton.

AgResearch chief information officer Phillip Lindsay and his team found a way to tunnel IP over X.25 to ensure both European and US connections. “We were at the leading edge in a lot of areas but we had a relatively small research effort compared to many overseas nations, and always had to leverage what we could. There was a lot of animal research going on in the US that was of interest to us. Our collaborators were using new tools and we had to be on board to be part of that.”

When AgResearch transitioned out of X.25 in the early 1990s and became more actively involved with Waikato, Lindsay and Houlker found themselves working long hours maintaining the beast that they had cobbled together. “John Houlker and I would both be at home late at night working on the network. I had an X.29 connection into our Cisco routers from my study and John would be on-line from his home resolving routing issues as well. When we first put the connections in we had quite a meshed network of the various parties, and everyone could see everyone else; if one party hadn’t done their routing properly, you would get traffic flowing across each other’s links. Otago University backhauled through our connection for ages before we knew what was happening. That was common in the early days and we had to pay for those links. Fortunately there was a good give and take approach, particularly among the universities and AgResearch.”

IP infiltrates parliament

Computer programmer Don Stokes had set up a bulletin board for DECUS in Wellington but when Government Print (GP) was turned into a SOE and readied for sale, the project’s sponsor was moved on and the site was never used. Stokes had plugged the system into the DSIR and continued to experiment with accessing news and email over the Internet but the DSIR wasn’t happy with him hogging its bandwidth.

Stokes had been experimenting with the email tools bundled with the Unix operating system and the X.25 messaging network to connect remote sites. Reliable email was still some way off and the addressing scheme was merely for identifying different machines. Alternative access through Victoria University gave Stokes an insight into how things were evolving beyond the proprietary environments, and he and others in the IT department at GPO began to experiment with the next level of communications. The Post Office had laid a 10Mbit/sec fibre cable[22] between GP’s Mulgrave Street offices[23] and the Hansard office in the old Parliament building, which was used for early experiments with the IP communications protocol. GP operated a VAX 11/750 at Hansard, which was used to prepare and transfer the text for the Hansard publications, for printing and insertion into parliamentary full-text databases.

Stokes had established IP addresses on GP’s internal VMS machines and parliament’s Data General AOS/VS systems and begun sending emails and transferring files between the two. “I was able to get our email from the outside world into Parliament using UUCP before anybody realised it was possible.” Internal Affairs, which looked after the parliamentary network, began to show interest as GP was already servicing the Dictionary of New Zealand Affairs Biography, which had an office in the corner of the historical branch.

An alternative mail system using DECnet had even greater reach within the parliamentary system. “You could actually route mail from outside, through our system and into Parliament, but I didn’t consider it wise to suggest that the prime minister could be sent email by anyone with Internet access,” said Stokes. There was some talk of connecting via IP to the Ministry of Education, and the Department of Internal Affairs which supported the parliamentary network and the National Archives. “It seemed to me GP was in a perfect position to drive development of an IPNetwork in the government sector.” However there was a strong belief among government types that the OSI protocols would replace IP, and, GP’s new owners were more interested in getting the core businesses right.

When Parliament Buildings was to be refurbished, Hansard moved to Bowen House, so GP requested Telecom move the fibre cable to the Beehive basement. “We could pick it up from there and get the connection repeated onto thick-wire Ethernet running through the tunnel under Bowen Street and up to Hansard.” Telecom responded that it knew nothing of this fibre, it had never done dark fibre before; its own fibre was always single mode, and anyway it only ever sold ‘services.’ Stokes assured the company that GP had two ends running 10Mbit/sec. Telecom came back: “The fastest service we offer is 2Mbit/sec ‘Megalink’…that must be five times 2Mbit/sec…over what length was that again?” The result was that a half-million-dollar bill turned up at GP for back rent of the cable. “Larry Hall, the Computer Bureau operations manager passed it around the office and we laughed a lot. Eventually we stopped trying to play ‘our minister is bigger than your minister.’”

As GP was still a government department, and the affair affected service to parliament itself, there was some pressure placed on Telecom, which was still an SOE, to be more reasonable. “The back charges were quietly forgotten, but the best deal Telecom was prepared to do wasn’t good enough. We installed a 10Mbit/sec digital microwave link from Mulgrave Street to the Beehive, which replaced the fibre at a cost of about $70,000. It worked well, except when cranes on the High Court construction site got parked right in the beam.”

The partially completed DECUS site was handed over to Richard Naylor at Wellington City Council’s CityNet. The GP Office became GP Print Ltd in 1990 and was sold to Graeme Hart’s Rank Group, which began focusing on electronically submitted print jobs on magnetic media, via email, dial-up file transfer, or direct connections. “To me, IP represented a solution to the mish-mash of protocols and systems in use for transferring information between ourselves, our clients and between co-operating agencies so everyone could attach to one big network. I began agitating within GP to try to make IP connections to other agencies but it wasn’t to be,” said Stokes. In late 1991 Victoria University advertised for a network systems manager, and Don Stokes was off to a role that would enable him much greater freedom to pursue the IP revolution.


[1] Marcia Russell, Revolution: New Zealand from fortress to free market, Auckland: Hodder Moa Beckett, 1996

[2] D. Henderson, Economic Reform: New Zealand in an international perspective, NZ Business Roundtable, 1996

[3] Alan Cameron and Claire Massey, The New Zealand Experiment – Has It Worked For SMEs?, New Zealand Centre for SME Research Management Systems Department, Massey University

[4] Milner had been responsible for establishing the second satellite dish at Warkworth in 1982, refurbishing the original Warkworth satellite station a year later and building the earth stations in Wellington and Rangiora

[5] LandWeber helped found the US university Computer Science Network (CSnet) and played a major role in the development of the global Internet. From 1982 LandWeber helped establish the first Internet gateways between the United States and countries in Europe, Asia, and Latin America, including co-operative relationships between CSnet and national network projects. His informal workshops led to the forming of International Networking Conference (INET), which since 1992 became the annual conference of the Internet Society. LandWeber has been president and chairman of the board of the Internet Society and its vice president for education. Source:

[6] Former head of the Federal Networking Council of the United States, who headed the NASA Scientific Internet

[7] Founder of Pacific Communications programme (PACCOM)

[8] Madin had built the West Coast Federal Internet Exchange and helped convince the US government to move entirely to TCP/IP rather than the OSI protocols it had been contemplating. Madin invented the concept of the Internet exchange, which inspired Houlker to introduce a similar exchange (NZIX) at Waikato in 1990 and later in Auckland

[9] Ping: A test of response time between your computer and the host your are trying to connect to, a rapid response time is desired, a slow response time suggests poor performance or issues on the line or with configuration

[10] Telnet is a user command and an underlying TCP/IP protocol for accessing remote computers, including requesting specific files. The result of a Tenet request would be an invitation to log on with a user ID and a prompt for a password

[11] Interview with Nevil Brownlee, March 2007

[12] Mark Davies, History of the Internet debate on NZNog newsgroup, March 2001

[13] Michael Newbery, History of the Internet debate on NZNog newsgroups, March 2001

[14] Mark Davies, History of the Internet debate on NZNog newsgroup, March 2001

[15] On 17 October 1989 a magnitude 7.1 earthquake struck the San Francisco Bay area, collapsing a section of the San Francisco–Oakland Bay Bridge. Damage was estimated at almost $3 billion in San Francisco, half the total damage figure for the entire earthquake zone. The earthquake knocked out power to San Francisco, and the city was dark for the first time since the 1906 earthquake and fire. Power was fully restored by 20 October. Emergency telephone service became sporadic because a fire broke out in the 911 telephone equipment room. The quake killed 62 people throughout Central California, injured 3757 and left more than 12,000 homeless:

[16] Computer programmer Kevin Robert Elz, also known as kre, was a pioneer in connecting Australia to the Internet and more recently was involved in getting Thailand, where he now lives, on line. He helped develop Internet-based research in Australia and operated the .au domain name from 1986 through to the late 1990s and managed the .aus Usenet hierarchy from the 1980s. Some of his achievements include developing a number of important Internet RFC documents, helping connect Australia to the worldwide Internet, developing the Internet-based research network within Australia, and operating the .au domain registry from 1986 through to the late 1990s. He also managed the aus.* Usenet hierarchy from its inception in the 1980s until the mid-1990s. Amongst undergraduate students within the University of Melbourne Computer Science Department in the mid-1990s, he was considered a semi-mystical figure, having written the BSD operating system’s quota system and contributed to its timezone management system. He was rumoured to have a work contract that allowed him time off whenever cricket was being televised. Source:

[17] Stone was a Masters student who later went to Stanford University to complete his PhD. Can be contacted via Professor Peter Barett, his stepfather, who’s in Earth Sciences at Victoria

[18] PCroute and PCbridge are software programs for IBM PC computers that can convert a PC with the necessary network cards into an IP router (PCroute) or an Ethernet bridge (PCbridge).

[19] Tony went on to work for Teletrend, which became Allied Telesyn

[20] Ace routers were developed by the DSIR’s Physical Sciences and the Information Technology Group, which became Industrial Research Ltd when the CRIs were formed in 1992. In January 1994 Network Dynamics Ltd acquired the network engineering team of IRL, the Ace Router and all rights from Industrial Research Ltd. It entered a partnership with Securicor 3Net in December 1994, which eventually purchased the company in 1996. In September 1997 Teltrend Inc acquired Securicor 3Net and the New Zealand operation became Teltrend (NZ) Limited. In turn it was acquired by Allied Telesis Japan, and became CentreCOM Systems before reverting to Allied Telesyn Research. The company now has a huge business supplying routers and network equipment worldwide with a 170-strong professional engineering division in Christchurch. Source:

[21] ECL subsequently become the country agents for Cisco then went through a series of transitions becoming Case, then Dowty and later Logical, which was purchased by IBM

[22] The cable was a multi-mode fibre, carrying Ethernet from the ‘thick-wire’ (10base5) cable in the Hansard office to that in the GPO machine room

[23] Now the National Archives building