Abstract: The disclosure herein describes establishing an endpoint within a network for a customer to reduce latency of the customer's network. Latency data associated with accessing a first endpoint within a network by end user devices from a plurality of geographic regions is collected, wherein the end user devices are associated with a plurality of customers. Performance scores for the geographic regions are calculated based on the latency data. A subset of the performance scores of one of the plurality of customers is selected. A second endpoint is identified within the network for the one of the plurality of customers based on the selected performance scores. The second endpoint is established for the one of the plurality customers for network traffic originating from the one or more of the geographic regions to reduce latency. The analysis of the network traffic data for customers improves network performance.

Abstract: A method for load balancing among multiple endpoint computing systems of a domain includes receiving a request to access the domain; identifying a target endpoint from the multiple endpoint computing systems based on routing logic for the domain; obtaining a load metric for the target endpoint; determining whether the load metric for the target endpoint satisfies an overload condition; and redirecting a fractional portion of traffic of the domain away from the target endpoint and to an alternate endpoint of the multiple endpoint computing systems in response to determining that the load metric for the target endpoint satisfies the overload condition.

Abstract: A method for load balancing among multiple endpoint computing systems of a domain includes receiving a request to access the domain; identifying a target endpoint from the multiple endpoint computing systems based on routing logic for the domain; obtaining a load metric for the target endpoint; determining whether the load metric for the target endpoint satisfies an overload condition; and redirecting a fractional portion of traffic of the domain away from the target endpoint and to an alternate endpoint of the multiple endpoint computing systems in response to determining that the load metric for the target endpoint satisfies the overload condition.

Abstract: The disclosure herein describes establishing an endpoint within a network for a customer to reduce latency of the customer's network. Latency data associated with accessing a first endpoint within a network by end user devices from a plurality of geographic regions is collected, wherein the end user devices are associated with a plurality of customers. Performance scores for the geographic regions are calculated based on the latency data. A subset of the performance scores of one of the plurality of customers is selected. A second endpoint is identified within the network for the one of the plurality of customers based on the selected performance scores. The second endpoint is established for the one of the plurality customers for network traffic originating from the one or more of the geographic regions to reduce latency. The analysis of the network traffic data for customers improves network performance.