Wildfalcon

Shortly after I started my PhD my supervisor sat me down, and told me something that has stuck with me ever since. He said that in a couple of years, if I was asked at a party “What is your PhD about?” I should by then, be able to answer. Two years seemed an absurdly long time to figure out what it was I was working on. But as time passed I realised how much sense it made. In research, the difficult part is often identifying the exact question to ask. Once you can ask the right question, answering it is often little more than the application of routine tools and techniques. Tools and techniques I was to learn in the next two years, while figuring out what question to ask.

Indeed, two years or so later, I did have an answer ready to that exact question, though naturally as I applied to tools to answering it, I found that some aspects where easier or harder than I had first thought, and so the question slowly changed. In fact it soon got to the point where without quite a lot of work on the exact topic, you wouldn’t even be able to understand the questions I was looking at. This is the second part of what my supervisor was getting at. How can you present your very involved work in just a few sentences, and managed to not totally bore the poor guy who just made the mistake of asking what your PhD is about.

What I was in fact looking at (just in case you were wondering) was the best way to manage programs on a distributed set of computers, that appear to the user as a single computer. Different programs want to use the different computers in different ways, meaning we want some thing (a decision making gizmo) to sit between the users and the computers, and run the programs. Thats what I looked at.

Naturally it become more complex than that though. It turns out, that the interesting question ends up being, what do you do when lots of users ask the computer(s) to run more programs than they can cope with. Clearly the system has to decide some programs don’t get to run. I spent the next year or two looking at solutions to that exact problem, eventually finding a solution that not only covered computer science, but maths, economics, and even a bit of political philosophy and biological evolution.

All my my work eventually got written up as my doctoral thesis, Scheduling in a Grid Environment Using High Level Policies.