During the Traffic Generator testing campaign, we investigated the current results from the Back-to-Back Frame Benchmark test as a possible way to explain the varying results between different generators. What we found instead, were:

• good stability of the phy2phy results over time and across hardware platforms
• several ways to improve the RFC 2544 methods
• an important dependency between Throughput and Back-to-Back Frame tests
• a correction factor to improve DUT buffer size measurement (the goal of these tests)
• limited support of this test among the traffic generators.

See https://wiki.opnfv.org/display/vsperf/Traffic+Generator+Testing#TrafficGeneratorTesting-AppendixB:Back2BackTestingTimeSeries(fromCI)

As a first step to learn from and resolve the issues identified, we have prepared an Internet Draft and submitted it to IETF BMWG:

https://tools.ietf.org/html/draft-morton-bmwg-b2b-frame-00

We seek support of these updated methods across VSPERF's traffic generators.