don at nz.net

My sanitised, approved for general consumption CV can be found here.There you can find a potted history of the places I’ve worked and what I’ve worked on, so you can believe that I do actually know what I’m talking about when you get me or my company to do stuff.

This is not that document.

Rather, this is my rather random, opinionated and egotistical list of stuff that I’ve done that I think is cool, significant or interesting. There will be some overlap with the aforementioned sanitised version, but this is where I get to blow my own trumpet (or at least, my own tin whistle) about stuff I thought of, played with, and in a few rare cases even deployed. Some of this is significant. Some of it utterly trivial.

Warning: may contain facts.

And it definitely contains opinions and recollections that may or may not reflect reality as it was, or reality as other people saw it.

Also, I’m going to use the word “invented” here. What this means is that I thought of and implemented whatever I claim to have “invented” as original work. It doesn’t mean I was necessarily the first to do so, or that it wasn’t bleeding obvious to someone working in the field, or that there wasn’t at least some external influence that led me to that invention. It does mean that I think I made the connection without an existing example of the the invention to work from.

So, without further ado, here’s some the stuff I’ve got up to in the quarter century or so I’ve been hanging out in the industry.

1985. Ran my first BBS

On the Waikato University first-year VAX/VMS system. Nasty hack. But fun. Lots more BBS-ing followed through the rest of the ’80s and early ’90s.

1986. Taught a computer to talk

And listen to touch-tone results. I built a prototype IVR system using a DECtalk DTC-01for doing credit card authorisations. It worked well. Later on, I and a colleague persuaded it to sing surprisingly well.

1986-1988. Watched an industry die

I am fortunate enough to have been witness, and even an active participant in both ends of the cycle of technology. In the 1980s, the technology on the rise was small, microprocessor based systems; PCs and under-the-desk time-sharing micros became available and cheap enough for small business to buy and run.

Big slow dinosaur computers on the end of slow leased lines just didn’t have the immediacy of having your data in your office where you could kick it. Commercial time-sharing bureaus were already finding they needed to look elsewhere for revenue when I joined Datacom Systems in 1986; when I left in early 1988 they were barely hanging on, as small systems encroached on their turf and the effects of the recession following the October ’87 share-market crash bit deep. A year later, the team of twenty-something programmers and analysts I’d worked in had gone, and the three computer rooms full of kit was down to the smallest one.

This was also the period when I learned about Fear Uncertainty and Doubt (especially as wielded by IBM), and perhaps more hopefully, how networks of computers could become so much more than the sum of their parts.

1988. Thought the Internet was a good idea

Now, I don’t claim to be unique in this, even then. But you do need to realise that in the late ’80s, the prevailing wisdom in the computer and telecommunications industries was that the Internet was an interesting overseas experiment, and soon what had been learned would be incorporated into a new set of protocols promulgated by the International Standards Organisation, as the Open Systems Interconnect, and sold by big computer companies for use on big telco networks. Being a little systems programmer in a Government department at the time, I couldn’t help but notice that, for example, I could get my “Internet” RFC 822 email from one side of the planet to the other in hours (this was before the Internet actually landed in NZ), while I couldn’t get X.400 email across the room, ever. The standards were slow coming (the OSI Virtual Terminal standard never fully materialised), and the software was either non-existent or expensive. Or both (which took a certain amount of bullshit on the part of the vendors).

It was obvious to me that Internet standards and software were more agile, more widely implemented and as a result, much cheaper. And that meant people who needed to get stuff done could just do it without writing a long business case and begging for funds.

Don’t ever underestimate the stuff that gets done underneath the radar.

In the mean time, I got the (dial-up) email and Usenet news feed going, and my first “Internet” email address, don@gp.govt.nz.

And at the same time, I had umpteen different “networking” technologies to deal with — X.25, dial-up, statmuxes, DECnet, LAT, Appletalk, uucp, and various other abominations. Don’t get me started about how I spent my 21st birthday in the bloody comms room trying to get file transfer using the IBM 2780 bisync to work between some dinosaur typesetting system and a VAX, over a 2400 bps, half-duplex synchronous modem link to New York via umpteen satellite hops.

It was really about time for some sanity in the networking field.

1989. Brought Parliamentary reporting into the modern age

The Hansard office used typewriters for presenting the record of members’ speeches to them to correct. Digital DECmate word processors were bought to replace them. Problem: the DECmate is designed to connect to a single printer, and a broken/jammed printer could completely upset the flow of work through the office. That doesn’t, on the surface, sound serious.

But Hansard works in real time; a member makes a speech in the House, and Hansard has a copy of the speech ready for correction within (I think) an hour. There’s no time to mess with printer jams or be wondering where in the queue your print job is. So I implemented a sort of “software switch” which would select the nearest free printer. Printers could be taken offline as well, and if all else failed, three of the DECmates had local connections (via hardware switch boxes) to a printer. It all worked remarkably well, and smoothed what was looking like a rocky transition from the typewriters.

Don’t underestimate what that last sentence means. Hansard is a high pressure place, employing staff with amazing secretarial skills. When Parliament was in session, people are running in and out, the division bells are going, everyone is banging away at keyboards. It’s stressful just being near the place. So the last thing anyone needed was the wheels falling off the technology …

1990. Brought email to our politicians

Really. Except they didn’t know it. We (the Government Printing Office) used a package called PMDF to do email routing via our UUCP-base email/news link. I also used PMDF to front-end a nasty hack to allow email to be sent from the GPO VMS systems to the Data General systems in Parliament Buildings. So it was only a small jump to setting up mail forwarding from outside all the way in. Did I make this jump? Oh yes.

Outbound mail didn’t work, due to addressing limitations in the DG end of the interface. But although nobody ever did (except in test), you could send mail from anywhere with Internet email to any user on the Parliamentary system. And that included our elected representatives. I’ll point out that this was long before anyone invented the term “spam”, and the Internet was a much smaller place … and you could walk into Parliament Buildings without even showing ID.

1990. Sent email to a fax machine

A PC, a GammaFax card, some nasty hacks with VMS to PC network (nasty in itself), and quite a bit of code. But it was quite a nice little email/text to fax gateway.

1990. Sent email to my home PC

And Usenet news. First using a news/email package called Waffle under MS-DOS, later running a (16 bit) 7th Edition Unix clone called Coherent, with uucp and mail/news software.

1991. Implemented Life in a text editor

That’s John Conway’s “game of life”. Stick that phrase into Google, you’ll soon learn more than I’m prepared to write here. I got my favourite text editor (EDT on VMS) to perform Life iterations on a game board stored as a text buffer. For no other reason than to prove it could be done.

1991-1992. Still thought the Internet was a good idea

All the good stuff to do with the Internet in Wellington was happening at Victoria University. And in 1991 the VUW Computing Services Centre advertised for a “Network Systems Manager”. How could I possibly pass that up? My attempts to get GP Print Ltd (the now privatised Government Printing Office) to connect to the Internet had not been successful, despite the obvious benefits to an organisation that took in a lot of electronic information.

The weekend before I started at the CSC, there had been a fire in the building, and in the chaos of the clean-up, nobody was offering to put a name plate on my office door. I printed my own, in the process giving myself a promotion by shortening my title to “Network Manager”. By the time it came to printing business cards, nobody thought anything of it.

1992. Launched NZ’s first ISP

The Victoria University of Wellington Internetworking Group, which later became NetLink was set up to sell leased-line Internet access, and built a business model to make this both affordable (for organisations at least) and sustainable. The model was so influential, that pretty much all significant ISPs copied our charging model in those first few years. It wasn’t until telephone companies started getting seriously involved that the volume model started to get replaced by per-dial-up-hour (which never worked for the “always on” connections that were NetLink’s bread and butter), and got replaced by data caps when early broadband deployments laid bare the shortcomings of per-hour charging …

Not bad for an “ISP” that literally started with a junk sale.

Of course the term “ISP” wasn’t used in 1992. I discovered some years later that what I’d been doing all this time years was being called an “industry.”

Now, my old friends at Actrix have always claimed the title of “NZ’s first ISP”, and of course, being founded in 1989, that claim has some validity. Although, I would point out that on that basis, Victoria University still wins, being founded a mere nine decades earlier. Actrix, in 1992, provided a news and email service (which Victoria’s Computer Science department was doing in 1985; that operation too was folded into NetLink) along with a few other things. But quite simply, the Internetworking Group was pushing packets for money as its primary, advertised business before anyone else in NZ was.

In 1992, NZ’s total Internet bandwidth out of the country was 64 kbps. That’s the bandwidth of one phone call, which tells you roughly how much the “traditional” phone companies (especially Telecom) cared.

1993. Put the Government on the Internet

Truth be told, credit for getting  our elected representatives properly on-line largely goes to Hamish MacEwan. But I couldn’t help but notice that what we implemented at one stage of getting Parliament onto the Internet, involving a DDS link to the Archives Building, which had a microwave link to Parliament Buildings and another DDS link to Internal Affairs, looked an awful lot like a drawing I did on the whiteboard at GP Print in 1991.

During that time-frame, the Internetworking Group put a significant number of Government departments on-line. Perhaps most significant of these was the Ministry of Commerce (now MED), where a recent hire to the newly formed IT Policy Unit, one Colin Jackson was getting interested in putting Government information on-line. (This was after he consigned the Government OSI Profile to the scrapheap of history, suggesting that it be replaced by a single-page public sector guideline “embodying pragmatism, responsiveness to user needs and to changing technology, and brevity.”)

I got a phone call from Colin one day, asking, “can I have www.govt.nz?” I had to think about that for a minute, before replying, “I can’t see why not.”

1994. Arrested a cracker

Well, there wasn’t an “arrest” per se, but there was a police car, seizures, interviews and prosecutions, albeit through the youth justice system.

Actually a couple of kids who thought running password crackers on their home PCs was fun. One was a nice kid who got dragged along by the other … who it turned out wasn’t averse to stealing from his friends.

Made the papers. More than once.

1995. Invented virtual web hosting

Back in the day, when the web was very young, all URLs looked like:

http://name-of-web-host.hosting-domain-name/customer-site-name/index.html

I thought, wouldn’t it be cool if you could say:

http://www.customer-domain-name/index.html

and was met with considerable derision from people who I otherwise rather respected, who said, “Domain names are just for host names. Soon everyone will use proper indexing systems and no-one will type in domain names on the web.”

Yeah, they really said that. I was, uh, sceptical; even if indexing and searching did become ubiquitous, I thought the shorthand provided by domain names would die hard. As of course proved to be true.

What if, I asked, we gave our web server a series of secondary IP addresses, and assigned one to each customer that we hosted? When a connection came in on an IP address, we’d know what web host it was associated with. I knocked up a proof of concept, which rather than actually serving pages, would perform a redirect into a traditionally hosted page.

By the middle of the year we (NetLink) had patched the NCSA web server (the forerunner to Apache) to modify the document directory based on the inbound IP address, thereby providing virtual web hosting more or less as we now understand it. (“Host:” headers, that allow virtual hosting without consuming an IP address per virtual host, were still in the future.) We packaged that up with virtual host email facilities, and made good coin selling that service. But I still have that proof-of-concept code, dated February 1995.

1995. Deployed NZ’s first national commercial Internet backbone

NetLink acquired the University of Auckland’s Internet operations, after an incident involving two of UofA’s customers which spooked the University’s administrators. Expansion into Christchurch and Dunedin followed.

1994. Created the iwi.nz domain

The story (largely promulgated by Jim Higgins) that the moderation policy for the iwi.nz domain consisted of, “Oh, I just ask my mother,” is not true. What is true is that I asked one Professor (later Dame) Evelyn Stokes, a geographer and historian at Waikato University and member of the Waitangi Tribunal, for advice before the domain was created.

1995. Put a volcanic eruption on the Internet

Video sound-bite version here. Full story here.

1996. Built NZ’s first wireless Internet service

Folk had been doing point-to-point connections using WaveLAN and Proxim “microwave oven band”  (2.4 GHz) gear for a several months. But at NetLink, we decided to set up point-to-multipoint base stations on top of the Cotton Building (and later, the Grand Chancellor Hotel in Christchurch, yes, the one with the drunken lean after the February earthquake) to provide Internet access to anyone prepared to pay the money. It was all a bit cheap and nasty, but it did give decent bandwidth (512 kbps per shared antenna) at a semi-decent price.

Of course the idea of using 2.4 GHz for communications over more than 100m (which wouldn’t have got off the University campus, let alone 2km across town) today is laughable, because that spectrum is now so crowded and noisy.

1996. Put a general election on the Internet

I’m sure this was Colin Jackson’s idea. I mostly greased wheels, organised the data circuit from the count at Avalon Studios to our web servers, and made the data transfer from the systems running the count to the web server work. Then played router jockey during the night. But it was almost a world first, and by far the most successful Internet coverage of a general election up to that time. It was also the largest web coverage of any sort (a whopping two megabits per second worth of data transfer at its peak) in NZ up to that point, hitting one million hits in 24 hours.

1997. Became an independent consultant

Because being told that my past work in building a visionary, multi-million dollar business was irrelevant, finding that the basic faith that I would share in the success of the enterprise I’d worked for years to build from nothing was in fact misplaced, that my life would be ruled by KPIs, and that a grey-haired fellow with very little practical experience across the table was my boss (and was paid 50% more) didn’t go down especially well. Being interrupted by said boss in the middle of a presentation about some cutting-edge work with virtual private networks with a comment that, “Visual BASIC was the way the industry was going,” made my departure inevitable.

I’m only a little bitter.

1998-2000. Tried to do some good

… as an Internet Society of New Zealand (ISOCNZ, now InternetNZ) Councillor. Served on SRS working group (see below), technical committee and convened the 0867 working group.

Successfully walked a very fine line between the SRS working group, working on the DNS for Domainz (sub-contracted to Glazier Systems), Council activities and my own interests and opinions.

1999-2000. Designed a DNS registry model that works

OK, I had help with this one. I was a member of the so-called “Hine Commission”, consisting  of Professor John Hine (Convener), Rick Shera, Steven Heath and myself, and the ultimate result was the DNC/NZRS business model for InternetNZ to carry out its NZ domain registration processes, known as the Shared Registry System.

I won’t go into the history of this (it was messy), but I will go on record as championing the following features for the SRS:

• Focus on providers of actual Internet services;
• Contact details in the SRS, so that there would always be an authoritative, externally accessible and auditable record of who owned a domain, independent of registrars;
• Password (UDAI) based transfer of domain names, allowing a registrant to authorise a registrar to “pull” a domain from an incumbent, rather than some arrangement requiring the authority of tat incumbent.
• DNS glue record objects contained within the enclosing domain, not separate objects as in many other registrars.

1999-2001. Built NZ’s first actual TLD DNS infrastructure.

Because, believe it or not, the NZ top-level domain didn’t actually have any. Oh, DNS service obviously happened, but it happened due to a bunch of grace-and-favour arrangements with various universities, telcos and even NASA. But actually having DNS infrastructure operated by or for the DNS registry and built for the purpose was a novelty in 2000. Oh, and the level of grace involved was staring to get a little low too.

2003. Built an ISP.

Again.

2004. Invented a way to save IPv4 address space

Because, unless you live in a cave, you’ll be aware that address space is a scarce resource. The technique is described here.

2007. Designed office space

We needed an office fit-out. The floor-plan had no right angles. I drew up a plan and gave it to the project manager, expecting it to get re-drawn “properly”. Instead, they just blew up my drawing and built from that. I was sure I’d stuffed up something that a professional would spot, but the only mistake requiring remedy was a doorway placed correctly on my drawing and just put in the wrong place.

2007. Built a data centre

Oh, we needed one of those, too. Since our previous lease had run out, there was a wee bit of pressure to get all this done on time.

Coming in on budget was a bonus.

2008. Designed and implemented a routing protocol.

The Fringe Routing Protocol. Because I like to be able to lie to my routers about the topology.

2009-2010. Showed how carrier layer 2 networks should look.

Architecture of the EnableETHERNET layer 2 network in Christchurch. Virtual circuits between a provider head-end, and customer ports. Simple, secure, reliable and manageable. And fast. Oh, and it appears to be earthquake-proof.

Also contributed to the TCF layer 2 interface specification for UFB.

2010. Became a web star

Or so I’m told. Because of this.