Given at Cloud East 2013 in Cambridge on 24th May 2013.
The Guardian Content-Api serves a little over 11 million requests per day, mostly from internal systems that depend on it to serve the main guardian website, the mobile site and other digital distribution systems.
We undertook a project to completely re-architect the API from a poll based system based on Solr in our data center, to a message queue based cloud service based upon Elastic Search, all requiring no downtime, and no missing content.
In this session I'll explain the reasons we decided to undertake the project, and the approaches we took to ensure it was delivered successfully and I'll explain how we decided upon the technology, and what issues we had along the way.