Global load balancing provides sites with content delivery capabilities from multiple physical locations. The motivations for doing so include: enhanced performance resulting from shortening the distance from the client to the server, additional service capacity by spreading the load across multiple servers, and enhanced availability resulting from the use of multiple servers.
The key ingredient for deploying global load balancing is a geodns enhanced dns service with the ability to answer queries based upon the geolocation/geoip characteristics of the client system.
While global load balancing is provided by geodns enhanced dns, high availability is achieved using failsafe server state monitoring. This capability causes the dns system to direct requests to the remaining, more distant servers in the event that the local servers are unavailable.
Maximum availability requires the deployment of a global load balance configuration targeting locally load balanced clusters. In that configuration the global load balance capability will continue to direct requests to the most suitable locally load balanced cluster until no locally load balanced servers are available.
A final enhancement is to disperse the locally load balanced servers across multiple data centers within a metropolitan area. This allows a local cluster to remain in service even if a specific data center housing a server in the local cluster is taken out of service.