Node.js at Scale
In 2013, the npm registry was the paradigm example of an overloaded and unreliable system. After 4 years of side-project volunteer support and massive growth, it was showing signs of strain. By the end of 2014, this critical component of the Node.js experience has become about as reliable as the air we breathe, with several months of 100% uptime. CJ Silverio, head of DevOps at npm, Inc., will discuss how we got from there to here, and the challenges involved in refactoring critical production infrastructure without downtime.
CJ Silverio, Director, Engineering, npm
From Node Summit 2015 (February 10, 2015)