Examining Our Availability Requirements

The architecture supporting the static content for www.example.com must maintain the same (or stricter) availability requirements as the dynamic systems because it is a necessary part of the overall service. Our peak traffic is expected to be 11,750 requests per second, so we should be able to satisfy that metric with four servers, leaving us two for truly unexpected traffic surges and hardware failures.

This means that we must build a two-fault tolerant system that has remedial balancing of traffic. Multiple DNS RR A records should be sufficient for distributing incoming requests naively across the six machines, leaving an IP availability issue that can easily be solved by Wackamole.