BTI Systems Blog

Measuring Latency: Get the facts (Part II)

Posted July 6, 2010 by Jason Smith

As a follow-on to my first note on measuring latency, another interesting piece of information is around test set resolution.  We use test sets from one of the industryleaders in this space and our manager of our System Verification Team has inquired directly to the manufacturer on the best resolution achievable at 1, 2.5, and 10 Gbps rates.  It’s currently in the 16-20 nanosecond range. To put this in perspective, it takes light 5 nanoseconds for light to transit a 1 meter fiber patch cable.

Our test sets provide 20 nanosecond resolution.  This means that a reading of zero nanoseconds could mean a value anywhere between 0 and 19ns, likewise a reading of 40 nanoseconds could represent a value of between 40 and 59ns.  Based on this when our test sets indicate an average latency reading of for instance 11ns we represent in our test reports as less than 20 nanoseconds (<20ns) based on the resolution available.

Ask how a vendor tests its latency, for sure the answer should be RFC2544 but second to that, what is there test set resolution as it will dictate their true latency measurement capabilities!