Jaxon Van Derbeken, writing for the SF Chronicle:
The website - run out of a single network server in Denver - normally handled about 19,000 visitors a day and cost $180,000 to set up. It began failing within two minutes of the 11:28 a.m. crash and went dark completely in 30 minutes, hit by a wave of as many as 75,000 users, Schuler said.
WTF? Someone paid $180,000 for a server to handle 19,000 visitors per day?
That’s 13 visitors per minute. A person could handle that traffic volume with a cell phone.
The airport’s 8-year-old website was “ancient” by current standards, Schuler said. On Aug. 20, the airport replaced it with a system that works off an Amazon cloud-based platform.
“The plan was to move the website to a system that did provide redundancy,” Schuler said, “so our response would be better than what happened on July 6.”
An ec2.tiny instance should be enough.