A few years ago the Guardian (guardian.co.uk) completed a major rebuild of it's systems that produced a single application that succesfully served the high volume of traffic to the site, accomplished many of the project's other goals and was judged a success. However, we weren't satisfied. Any new addition risked affecting the whole site's performance and stability, the code base was continuously growing and there was little room for innovation and experimentation.
In this talk I'll discuss the various ways we've attempted to break down the site into a collection of smaller applications including the difficulties and missteps as well as the successes.
Phil is a software architect at the Guardian who has spent long enough eating his own dog food to realise that it's not always someone else's incompetence that leads to an imperfect solution. Phil firmly believes that architects should regularly get their fingers dirty in code and hacks some Scala whenever he gets the chacne.