In my first article, part of a three-part series, published in the launch issue of the InterGlobix magazine, I described how the London Internet Exchange (LINX) came into existence in 1994, as a place for ISPs to exchange their traffic.
LINX survived a real crisis in the year 2000, and just after that is where I come into the story.
There had been quite a dramatic change in both the Board and staff of LINX, and the Board then recruited me as their new CEO. I’ll pick up the story after we survived the turbulent couple of years that immediately followed, which I had covered in my previous article.
So, membership confidence was back, but there were still plenty of challenges to address.
Finding/replacing and keeping good staff was an immediate one, and I could probably write a book about how we addressed that challenge. The simple version of it is that we came to realise that we couldn’t compete head on with the networking industry, especially in London – so we needed to emphasise our virtues. We had two that served very well. One was an historical accident, in that our headquarters had been set up well outside of London, in a city called Peterborough. The cost of living and hence salaries there are much lower than in London, and office accommodation (and all the related costs and taxes) is a great deal cheaper than London. So that kept our costs down, and we only employed key staff in London rather than Peterborough where it was really necessary (essentially the core senior engineering team). The second was the realisation that for a core Internet infrastructure company, we cared much more about quality of work than we did about rigid office hours and where you are working. This evolved into an approach to flexibility which our staff value highly to this day. Flexibility can take many forms, but from an employer’s perspective you get a wonderful payoff from the responsibility and approach taken by your staff, when you show that you also respect their need for flexibility, not just about the hours of work, but in how they blend their work and home lives to their satisfaction.
Another immediate challenge was the switching equipment that we used to provide our peering LAN. We were somewhat constrained, not only by what we could afford, but also by the somewhat limited range of offerings on the market. This took us down some interesting routes, which will dominate the next part of my story.
From the very earliest days, LINX struggled to identify suitable equipment to run the exchange. There were many reasons for this, not least being the volume and heterogeneous nature of traffic being passed across it. We were somewhat feature averse, since complexity tended to be the enemy of reliability and robustness – but this was against the trend of what vendors thought the market wanted. Switch software tended to be somewhat flaky, and as LINX grew to be one of the very largest exchange points (in terms of peak traffic and connected ports), any flaws in either the hardware of software tended to be discovered very quickly – and inevitably, this sometimes happened in the production environment.
One of the main impacts of this was to force us to put enormous energy and ingenuity into equipment qualification – quite a challenge for such a small company. Ironically, the switch vendors seemed to appreciate this, even though it could never be as effective as we would have liked, since their own approach to design qualification and quality assurance was also compromised by their finite available resources. There is another interesting back story about all of this which I might write another day…