Unlike other applications such as email and web browsing which have some tolerance for packet loss, latency and delay variation (jitter), with Voice over IP (VoIP) calls there is only one chance to get the packet through in time for it to make any sense on the other end.  Packet loss, high latency and jitter on networks supporting VoIP can make communicating awkward or even unintelligible in extreme cases.

Because of it’s “just in time” nature, VoIP as a network application is particularly susceptible to transmission delays or inconsistent access to media.  Because of this sensitivity network vendors and design engineers have developed methods of classifying real time protocols such as VoIP and video teleconferencing, and moving them to the front of the line in order to minimize any disruptions to the transmission stream.  These protocols along with policy guidelines are known collectively as Quality of Service (QoS)

The most popular method for providing QoS in data/voice networks is DiffServ (Differentiated Services), whereby traffic is classified based upon tunable criteria, and then sent out on the network in the desired order.  The mechanisms used to tag packets are shown to the right, and various classifications are shown below.





Check out this excellent whitepaper from industry leader Cisco Systems entitled “Diffserv – The Scalable End-to-End QoS Model” for more detailed information.