A single server is a single point of failure. For web services where downtime is expensive, we design and build architectures that distribute load across multiple application servers, automatically take unhealthy servers out of rotation, replicate the database and can fail over to a second location when needed.