RTT and availability. mostly at DOE/HEP (ESnet) sites.
Don mostly uses homegrown stuff to measure end-to-end loss, and he analyzes
tcp traces. Don wants a NIMI where he can schedule experiments. At NIST
(Kevin Mills), pingER is used in a one-way measurement infrastructure(?)
Commercial Performance Measurement Companies
netletts
Uses own code for measuring end-to-end delay versus message size of tcp/udp
streams. It's limited: only trying to analyze probabilistic QoS. Takes
higher level stats: (jitter), delay, loss). Uses lighter weight version of
tool from inside a protocol (ref:shiv).
Cooperative Association for Internet Data Analysis (CAIDA)