Global Server Load Balancing. Load balancing refers to distributing the workload optimally across Servers.
In GSLB a website is hosted on servers located at different locations.
If a user in the US browsed website.de in usual webhosting he would be served by the german server only. In a GSLB architecture he will be served by a server in the US, if such is present.
The archetypical technique to do that uses a Master Server and Slave Servers in different locations. Whenever a request is raised, the Master Server checks the location from where it was initiated. Then cookies it redirects the request to the nearest server location.