Trip report for Cambridge Geek Night 5 (#cgn5) at The Cambridge Union.

See also on Lanyard.

Topics included open data, scaling services, and startups.

The talks:

  • Lisa Evans (@objectgroup) from the Open Knowledge Foundation talked about open government data for "Where Does My Money Go?" Nice visualisation of aggregate and granular data, and war stories of obtaining data from government and dealing with data compatibility issues.

  • Martin Kleppman (@martinkl) from Rapportive. Rapportive is a value-add for Gmail, otaining info related to items in your inbox by aggregating external data from social networks etc. It is implemented as a plugin which inserts a JavaScript tag, which then loads the JS logic from Rapportive, which then does the fetching and DOM manipulation. Rapportive uses Nginx, Ruby, RabbitMQ, Redis, S3, github, heroku. They will be looking at Hadoop, Solr, Memcached. They outsource some tech expertise. And they're moving to San Fransisco. I asked a question about risks of reliance on gmail's page structure and whether bizdev contract were in place with Google. They address this by lots of checks so they can partially degrade the functionality, and aggressive monitoring. So far there have not been any changes that affected them. The business model is freemium, with premium chargeable features in the pipeline, though one has to wonder if a Google buyout is the real target. I've heard their UI replaces the ads inside Gmail; wonder if Google will take action against that. Popular quote from the talk was "Scaling calcifies: don't do it too early" referring to the problem getting locked into a specific scaling technology too early.

  • RedGate sponsored the bar. Their intent was to do some hiring, so they spent some time explaining why their company has a great culture and would be fun to work for. They handed out some swag, including copies of The Book of Red Gate 2010 which I liked quite a bit, to my surprise. I couldn't resist doing the Clive Code Challenge, where I didn't do too badly.

  • Someone mentioned a Clojure Cambridge User Group meeting Tuesday 5th October 7pm in The Punter. I'd be interested in the scaling aspects, but I may just not be Lisp enough.

  • IdeaSpace (@ideaspace) is a co-working space and networking community at West Cambridge Research & Development Park (not far from the MSRC office). £50/month membership. Friday afternoon they have a CamJelly, a public/social/meet thing. Should check it out if I find myself in the area.

  • Mat Clayton (@MatClayton) from MixCloud gave a talk about scaling an audio service: users upload/download DJ-style mixes. Fascinating and entertaining. Their main problem was not the scaling user data/access but problems with bandwidth costs, which at times threatened the company's survival. They started off on EC2, moved content service to, then downgraded bitrates (from sometimes insanely high), and are now on OVH because they're cheap. Migrating massive amounts of data between providiers alone is a challenge. Various issues around music licensing were discussed. For UK purposes they're a radio station, and have appropriate PRS for Music licenses with track reporting. Mat works out of Cambridge, near the redgate folks, the rest of the company is in London. They managed without VC funding -- impressive.

  • Michael Brunton-Spall (@bruntonspall) who works at The Guardian plugged UK Scale Camp.

Met various new people:

Meta comments: Some parallels with the OpenTech 2010 conference a couple of weeks earlier: Lanyard is popular, people ID themselves with their twitter username these days, and live event twittering via hashtags is common and useful.