If a server fails, that server is either automatically replaced by hot-spare or the server is removed from the cluster and it's work is handled by it's peers.
Each of the duties in the hosting environment is shared across no less than two servers. If one server is busy, one of it's less busy peer servers will be assigned the task and respond instead.
Our physical hosting environment is secured by two pass card checkpoints. Our network is only accessible from the Internet through redundant firewalls. We block all traffic except only what is required to provide your services. We only run the most secure, industry standard protocols.
Off-site servers run daily backups of all files, databases, and configuration.
The entire hosting stack is monitored 24/7/365. Should any part of the system fail, our team is notified within a minute or two by email and by text message to our mobile devices.
All servers are installed with the Linux operating system and only the bare essential services to maximize efficiency. The average up-time using this platform is measured in years instead of weeks as compared to the alternatives.
All servers and supporting appliances are instantly switched to battery backup should the power fail. And, should the batteries run out, diesel generators take over. The server room is cooled 24/7/365 and filtered for dust.