The Brief
We hesitated for two years before publishing this case study. Fyre Festival, conceived by Billy McFarland and Ja Rule to promote the Fyre app and planned for the Bahamas in April and May 2017, collapsed famously and publicly, leaving attendees stranded and suppliers, ourselves included, writing off unpaid invoices. McFarland was later sentenced to six years for wire fraud with a $26 million forfeiture, and the Netflix and Hulu documentaries tell that story. Ours is the technical one: the website held.
In partnership with Matte, we built the festival's responsive website with third-party ticketing and payments integrated via API, which demanded frequent updates and cache purges to keep ticket availability accurate. With over a million visitors an hour expected after launch, driven by influencer and global media promotion, we designed for genuine scale on Amazon Web Services. The largely static site leaned on extensive caching, with CloudFront distributing content globally for fast load times, while the demand for HD full-screen video raised the bar further. In AWS's US-EAST-1 region, Elastic Load Balancers spread traffic across multiple EC2 web server instances that auto-scaled with load, backed by a Multi-AZ database cluster, with imagery served through CloudFront from S3.
At launch the site handled over 6 million visitors in the first three hours without any downtime, traffic on the order of Glastonbury or Coachella.
Whatever went wrong on the island, the engineering held, and capacity planning of that order now underpins every platform Node operates.