This was a pretty rambling session lead by Julian Simpson. His site contains lots of build and deployment related patterns and best practice. Unfortunately I didn’t find this session helpful but picked up the following:
- Make deployment an atomic operation – makes sense, I’m thinking tricky with Alfresco but doable.
- Pull upgrades, don’t push them – avoids the “node X is down” problem, i.e. it will therefore be out of step when it’s brought back up.
- If we ever do any Government Gateway work, they provide a VM that gives you a reference architecture – quite a neat idea.
- Write deployment scripts really early in the process. If it’s risky, maybe even before any code.
A BoF session followed the above where a load of very passionate developers brainstormed ideas around saving lots of tax payers money by avoiding the very public IT disasters we’ve had over the years, e.g. NHS “super computer” cancelled at a cost of £12.7 billion just before Christmas. That’s a lot of money. This resulted in a commitment to do something about it as a group – I joined and will let you know what happens next. I then did some networking, had a couple of beers and some food and then joined a walking tour of the Strand area. This was followed by some more networking where I was convinced we should look at easyb for testing, Graddle as additional technology to our automated build and test suite and Capistrano to improve deployment – maybe but let’s get the most out of what we’ve got first.
