NetPath started as something we were fiddling with in the lab. We came up with some really cool tech tidbits like:
- Discovering the entire, complex path that connects users to services far, far away on the Internet. Often dozens and sometimes hundreds of routers, links, and servers.
- Quantifying the performance of every node and every link.
- Correlating hop by hop performance on a node or link that is used for a portion of the traffic to the end-to-end performance experienced by users.
- Determining what portion of latency is healthy and what portion is unhealthy. That way, we can correctly mark a link between two routers in the same building as unhealthy when it takes 7ms and mark a transcontinental link that takes 30ms as healthy.
- Reliably mimicking application traffic so we are allowed through firewalls and treated with QOS just like application traffic is.
