Network latency measurement and analysis system

Abstract

Deploying a point of presence (PoP) changes traffic flow to a cloud service provider. To determine if the PoP improves the performance of a cloud service to a client, actual network latencies between the client and the cloud service are measured. In more complex scenarios, multiple PoPs are used. The client sends multiple requests for the same content to the cloud provider. The requests are sent via different routes. The cloud provider serves the requests and collates the latency information. Based on the latency information, a route for a future request is selected, resources are allocated, or a user interface is presented. The process of determining the latency for content delivered by different routes may be repeated for content of different sizes. A future request is routed along the network path that provides the lowest latency for the data being requested.