Sometimes I get article proposals and then there's no follow up. Though these TripAdvisor data points are from 2010, I thought them worth sharing:
Our site serves in excess of 100M dynamically generated page view a day (all media and static content goes through CDN), and we do this with about 100 machines, no single point of failure, supported by distributed service architecture that that responds to over 2B requests a day, and a data warehouse of over 20TB that is used to drive email campaigns, SEM, and general reporting. We are a Linux/Java/Apache/Tomcat/Postgres/Lucene shop, and have built our own distributed computing architecture. We also maintain duplicate data centers (one active, one standby) for redundancy and maintenance purposes.
- A new twist on "data-driven site" by Mac Slocum. How a billion points of app data shape TripAdvisor's website.