I note the information in the following kb article about QOS Reflectors: https://helpdesk.paessler.com/en/support/solutions/articles/76000063284-how-can-i-monitor-qos-roundtrips-without-using-remote-probes

and have the following questions:

1. In a test setup, I installed a QOS sensor on a PRTG server, and installed Python for Windows on a Windows Vista machine. I then executed the reflector script on the Vista machine - and have been receiving results. Unlike the documentation, I have the "QOS TARGET" setting on the sensor set to "PRTG Probe" (where the documentation says use "Custom Target".

This setup is successfully capturing statistics - but my question is - do these settings in combination with a Windows machine running the QOS Reflector render the results invalid ? (FYI - they APPEAR correct - but I have no way to validate).

2. In attempting to setup a QOS Two Way sensor with "QOS TARGET" set to "Custom Target" per the instructions - I received the following error message:

THERE WAS A PROBLEM...

PRTG Network Monitor has encountered a problem. Your last request could not be processed properly.

INTERNAL ERROR: '' IS NOT A VALID INTEGER VALUE

Error URL: /addsensor5.htm?errorintervalsdown_=1&id=2102&inherittriggers=1&interval_=60|60 seconds&intervalgroup=0&intervalgroup=1&name_=QoS 2&packetcount_=1000&packetinterval_=20&packetsize_=172&parenttags_=&priority_=3&qostargetmaster_=custom&sensortype=qosroundtrip&tags_=qossensor&targetid_=1|Local probe (127.0.0.1)|127.0.0.1|&targetip_=192.168.1.98&targetport_=50000&timeout_=60&

3. In the results captured (per question 1), I note that both Jitter and PDV statistics are captured, yet there is a significant difference between (for example) Jitter (Max) and Packet Delay Variation (Max). Can you explain this and provide information on the measurement method for each of these ?

Your assistance is appreciated,

Glenn.

Article Comments

Hi,
please find my answers below.
1: Yes, the metrics are correct. There might be some differences between Linux and Windows because of the TCP Stack but this is a general thing and not limited to the QoS Reflector.
2: This is a bug unfortunately which will be fixed with one of the upcoming versions.
3: Actually Jitter and PDV are calculated differently. The Jitter is calculated according to RFC3550 and PDV is a bit more complicated as there actually two calculations, first we search each packet's following packet and do a variance analysis here. After the packets are received we do the same again and from the difference the PDV is calculated.
Best regards


Jun, 2016 - Permalink