Wireshark-bugs: [Wireshark-bugs] [Bug 10245] Measurement of response time elements

Date: Sun, 10 Aug 2014 13:25:27 +0000

Comment # 5 on bug 10245 from
Hi Evan,

Thanks for taking the time to review this.  Yes, you are quite right that
because we are not using client side and service side traces to calculate
spread values there is are inaccuracies, and I take your point that the
inaccuracy is more significant with higher network latencies.

In our experience in most cases these inaccuracies are still less significant
than other delays, particularly if we are investigating intermittent
performance problems.  However, I think we should at very least cover this in
the TRANSUM Manual and so I'll get this updated.

Further to your point, in reality we should never have a zero spread value as
even a single packet Request APDU or Response APDU would be delayed by one-way
network latency.  I'll discuss with colleagues possible ways to a) estimate
network latency or at least RTT, and b) make appropriate adjustments to the RTE
values depending on the client-side or service-side setting.

Best regards...Paul


You are receiving this mail because:
  • You are watching all bug changes.