Description
Network Resources when have a defined latitude and longitude are eligible for traffic routing via "Geographic Nearest". The redirector system for determining the nearest network resource is determined solely by distance. Essentially, the system will geo-locate a user's IP address, then also geo-locate the ip addresses of the network resources in the list of delivery options. Using a modified Haversine formula, it is determined which resource is closest to the user.
However, there is a "tolerance" factor, which means that the user can be deemed "far enough away" that both resources are essentially the same distance. An example would be if I am in Washington DC, but have nodes in Seattle (2,764.6 mi) and Los Angeles (2,668.4 mi). The difference of 96.2 mi is not relevant. At that point, the user will "round robin" between the two.
Round robin is not a forced function (state is not stored by the redirector), it essentially means "equal weight" of the list of resources in it. So between 2 resources, the long-term distribution will be 50%, but the short term might see runs of one or the other.
The geo-closest tolerance is roughly 5%, however that number is tunable by our operations team, in case they see too much load on one server.
In the attachment below, IP Addresses located in the Green circles will locate to the node within that circle. IP Addresses located in the red will round-robin between the two.
