A converged network is a network that transports not only data but also voice and video. Unlike data which is generally delivered in a best effort fashion, voice and video traffic need special care.
When data, voice and video share the same network, they are sharing the media bandwidth. This situation may lead to a lack of bandwidth, whose impact on regular data applications is not as hard as it is on voice and video applications. In fact, we can tolerate some delay in email delivery or a server response. But, to hear a choppy sound or a video conference where sound and picture are asynchronous is not appreciated.
Talking about bandwidth, we need to distinguish maximum bandwidth and available bandwidth. Maximum bandwidth, on a network path, is equal to the bandwidth of the slowest link. Available bandwidth of a network path is equal to the maximum bandwidth divided by the number of flows. Why? because there are many flows that traverse the network links, and each flow needs its own bandwidth.
Multimedia applications and critical data applications are sensitive to the following parameters:
- delay: is the time it takes for a packet to go from a sender to a receiver. It is also called end-to-end delay. At each hop of the network path, we have :
- fixed delays: serialization delays and propagation delays
- variable delays: processing delays and queuing delays
- jitter: is the variation (or variance, as Cisco calls it) of delay for a single flow. Remember that there are many flow traversing a network path and each flow is characterized by its own jitter value
- packet loss: this is the most obvious one. Packets are lost in one of the following situations:
- Tail drop: packets are dropped from the output queue of the interface
- Overrun: the CPU can not provide buffer space to the packet
- frame errors: when the system hardware detects errors such as CRC, runts or giants
- Input queue drop: may occur when the CPU utilization is high