Skip to Main Content
Today, standards-based quality of service mechanisms are not universally deployed across Internet service provider backbones. As a result, many carriers that are deploying VoIP networks are segregating voice traffic from the data and installing pure VoIP backbones in order to better guarantee high quality. Of particular interest is maintaining end-to-end delivery of the voice information within some time bounds. This paper examines sources of end-to-end delays associated with VoIP calls, discusses differences between fixed and variable rate coding, and illustrates how trading off packet size versus router delay can impact the number of voice calls supportable by the network.