This is a solution I recently deployed for a client.
There are two parts:
1) The flow of incoming/outgoing traffic: client initiates request and is directed to a server based on DNS interogation (reports show a 50-50% connection balance for two servers) and the HTTP server delivers the content;
2) Background service communication and data synchronization over the Internet.
Automatic file system replication ensures that periodic data synchronization scripts are unneeded; databases are automatically replicated and can be also load balanced; sessions replication ensures that no matter where the client lands the second time he initiates a request, his activity is not interrupted.
Notes: The system can be grown to larger sets.
Advantages: No Single Point of Failure. If one node fails, you just exclude it's DNS entry. If one hosting provider fails, you have another one available.
RedHat Certified Engineer (RHCE) with 8+ years experience as Linux admin. I have worked in various environments, from enterprise to medium/large, and now to freelancing and consulting based projects.
My experience covers a broad spectrum of the Linux administration domain, from basic servicing to web clusters. I have extensive experience installing and managing Red Hat/CentOS, Debian/Ubuntu and Gentoo servers.
I can set up high concurrency, high availability/load balanc...