Yes, the 100ms was totally in error. The product actually did 0ms on success and 50ms on error. We can assume 50ms for all throughput discussions to guarantee average throughput.

We have no wait until a failure, then 50ms is used until successful. I was doin... Login to read more.