As data passes through the internet, one piece of information is particularly challenging to manage; voice.
Part of its sensitivity comes from the timeliness of the data. Once a packet of data is delayed by over 50ms, then it is useless to a call and needs to get dropped. Often a packet can arrive late which introduces jitter. Even though the call statistics will show that the packet arrived, it was useless given the delay. Most Internet packets use TCP/IP (Transmission Control Protocol / Internet Protocol), the transport layer of this protocol asks the host to resend a packet of data if the receiver didn’t get the packet. For most applications, this happens so quickly that it is unnoticed. So streaming services use buffering to allow for this to happen; this is not possible with voice. There is no point in getting the “Hello” towards the end of the sentence.
That’s why measuring jitter by using voice packets is a sensitive method to see how a network or a connection to the internet is performing.
Voice also has as light as possible packet overhead. The Physical, Data Link, Network, and Transport layers are kept as light as possible to speed up the packets of data. We are improving the sensitivity of voice as a measure of a networks speed. We use the Jitter on a network as the measure of the efficiency of the network. Jitter is the variation in latency of packets of data and is measured in the time it takes for the packet latency to get across a network. By sending voice over a network to our servers and back again, we can see the delays that occur and therefore make inferences on the network’s efficiency. This gives a real-time measure of network performance.
Stressing the network
By stressing the network with injections of additional traffic, we can see the effect of that stress on jitter and packet loss. We don’t try to fill up a connection, see how it performs with multiple applications running. A network engineer can introduce their stresses on a network. They can simulate the maximum strain they would have on a network and then use voice over the top of this as a measure of the performance. Thus, this allows network engineers to shape or police their network.
