QCon London 2010
It's conference season again and this year I'm lucky enough to be speaking at QCon London! Okay, I kind of cheated since Nokia is one of the sponsors, but, nevertheless. My talk is about "Scaling to billions of people and places" in the context of Nokia Maps and location based services. The crowd was small but so was the room so it felt very personal and it was nice to be able to interact and answer questions from the audience in that format.
Thanks to those that came, here's the presentation along with notes for you to follow along with.
Ignite Berlin video
Sabine and I recently did a talk at Ignite Berlin on Lego Manhunt. Here's the result!
Ignite Berlin
Forgot to post about this, but Ignite Berlin is tonight! Sabine and I are giving a short talk on our little idea/project, legomanhunt.com.
Log analysis with Pig and gnuplot
There are lots of easy ways to analyse standard Apache (and Tomcat) access logs. Tools like Webalizer and AWStats have existed for some time and are well tailored to logs generated from "standard" web traffic. What happens when you want to get a bit more advanced, look for different trends and graph things that standard tools just won't do? When your servers generate gigabytes (or terabytes) of data, the natural choice right now would be to gravitate towards Hadoop. However starting to hack away at MapReduce jobs in Java somehow seems a bit, well, wrong to be honest. Then came along a little Pig.
Hadoop Installation Quick Start
We've just finished setting up a small 15-node Hadoop cluster at work and more and more things keep coming to my mind that could be solved or dealt with easily in Hadoop. However, relying on a production cluster to mess around with just isn't worth all the hassle (amongst other things). So with all of the various ways to actually get up and running with Hadoop, what's the easiest or best choice? Here are some options that are available and some comments on the ones that I've played around with the last couple of months.
