I am used to the term ICMP Ping Packet Loss but recently, I came across this other term UDP Jitter Packet loss
I will like to know the difference between the two, and of what importance it is to monitor one or the other.
Thanks in advance
1
-
Did any answer help you? If so, you should accept the answer so that the question doesn't keep popping up forever, looking for an answer. Alternatively, you could provide and accept your own answer.– Ron Maupin ♦Commented Aug 14, 2017 at 2:47
Add a comment
|
2 Answers
It depends on what protocols your applications use. Any protocol can have packet loss, and some packet loss is on purpose, e.g. RED randomly drops packets from queues to prevent TCP global synchronization (a bad thing). Also, when using QoS, you may want to police certain protocols over a certain bandwidth, and this drops traffic in excess of a specified bandwidth. A lot of loss occurs due to congestion, usually due to bandwidth over-subscription. It can also occur due to network attacks.
Jitter is different than packet loss. Jitter is a variation in the delay of packet delivery. It is important to minimize jitter in real-time protocols. For instance, VoIP can withstand a fair amount of delay, but even when you have a low delay, variations in the delay can cause big problems.
answered Oct 19, 2016 at 15:02
-
You wrote "Jitter is a variation in the delay of packet delivery" Does that suggest it is the tautology to say Jitter Packet Loss?– NTDCommented Oct 19, 2016 at 15:12
-
Jitter and Packet Loss are indeed two unrelated network metrics, and "Jitter Packet Loss" has no meaning.
However, "UDP Jitter Packet Loss" does mean something, when taken in the context of a Cisco IP SLA "UDP Jitter" operation, which among other things can measure per-direction packet loss.
Having per-direction measurements is very useful, beyond that it's hard to say which is more "precise" : packet loss, like throughput, is not a directly measurable quantity, so you are approximating. You take a set of packets sent over a period of time, and count those that didn't make it, but there can be any number of reasons why that set of packets is not fully representative of the total traffic : QoS is being applied, the interval between two packets in the set is very high...
In any event, having some means of measuring packet loss is a good thing, and typically an IP SLA operation will have been setup in a more "relevant" way than some random ping.
