Real-time multi-media services such as IP Television (IPTV) and video conferencing place difficult demands on IP networks. IP networks are inherently good at delivering messages on a best effort basis most of the time. But a late packet in a real time media stream is as good as a dropped packet. Everybody here is probably familiar with using queues to prioritize traffic and rate limiters or policers to slow down non-priority traffic. But as high bandwidth multi-media services become more popular, the highest priority queues become full, and even high priority traffic effectively gets best effort treatment. It is not cost effective to scale your network for peak capacity.
Fortunately, there are ways to gracefully handle peak periods and honor service level commitments.
True service assurance requires bandwidth admission control. If somebody starts watching a real time TV broadcast, nothing is more frustrating than having the show start out fine but then degrade and fail in the middle. If this happens one too many times, you either call customer service or switch providers. A good bandwidth admission control system, such as the SRC Policy Manager, reserves the bandwidth for the duration of the show.
Network enforcement policies must be dynamically provisioned in the network to ensure that the multi-media session has access to the bandwidth promised by the reservation. This assures that the multi-media experience will be assured for the duration of the show. And during peak times, it is important to have graceful messaging to the viewer. If insufficient bandwidth is available, the viewer should receive an intelligent message and offered options. For example, if an HD stream is requested but only SD bandwidth remains, apologies and the SD version of the content should be offered.
How do you model the bandwidth in the admission control system? With congestion points. Sound complicated? It doesn’t have to be. I’ll describe how easy it can be in the next post in this series.