Video applications can be very sensitive to aspects of network performance. After the video has been packetized for IP transport, the network’s contribution to the video stream’s performance is generally limited to delay, jitter, and loss.
Delays arise from physical limits (speed of light) as well as queuing mechanisms in routers and gateways that the packets traverse along the way. When delay increases above 400 milliseconds (camera to display), people become aware of it and the delay starts to impede interactive communications.
Jitter is the variability of delay. Buffers can be used to smooth out variations in delay. However, too much buffering adds delay and prevents effective interactive video.
Loss is particularly interesting for video applications in that loss (or when the packets don’t arrive in time) is highly visible. Due to the nature of video compression, a single packet loss can result in multi-second artifacts visible on the screen. Loss may be caused by poor links (last mile access link generally, or wireless links), a routing change within the network or many other unplanned outages.
Multiple network paths may also be available for a media stream to traverse to its destination. Some of these paths may be better suited for the video application than others. A performance-aware network can match the application performance requirements to the best available network path. Again, this process optimizes the quality of experience by minimizing jitter, loss, and delay.
Network management tools can help in making and analyzing the network for loss, latency and jitter. The baselining and subsequent comparisons of measurements against the baseline can help the network operator understand the performance of both the network and the delivered application video. Finally, understanding the deviance (fault isolation) from the baseline can help in reducing the time to fix issues.