The deployment of network video applications presents several challenges for the network and application administrator. These challenges can be categorized into having predictability, offering performance, and delivering quality.
The deployment and integration of voice into the network was relatively simple compared to the challenges being brought by multi-faceted IP video applications. As with Voice over IP (VoIP), video over IP allows the reuse and convergence of communications infrastructure. Where different types of delivery mechanisms (satellite, DVDs, taps, and coax) were needed for the various types of video, a single transport system can now be used. With VoIP, there was concern regarding some aspects of network characteristics such as delay and jitter. If the video application is an interactive one like video conferencing, then delay and jitter do remain important. However, in other types of video applications (for example video surveillance), the amount of raw bandwidth to deliver high quality video can also be a limiting factor. Additionally, unlike voice, minute network degradations can result in easily noticeable impairments that remain on screen for a longer amount of time.
Video can consume a great deal of bandwidth. With the High Definition (HD) video becoming more common, this usage amount will multiply quickly. Streaming a single hour of HD video (8 Mbps) with a modern video codec can easily consume more than 3.5 gigabytes of bandwidth on the network. Using earlier codecs, the value is twice as high. If the session needs to be recorded (for example, if it was an interactive meeting or an employee-directed broadcast), then you need more than 3.5 gigabytes of storage to save the content. Given the large numbers involved for a successful video experience, predictability of success must be assured.
The endpoint and the network may not have the capacity to deliver an hour of HD video to all consumers. Portions of the network may not have enough bandwidth to support both the high-quality video and regular business transactions. These contentions need to be controllable as well as predictable. Given business policies, a disruption of either the video or business transactions may actually be acceptable — but the outcome of the contention must be deterministic. Because of the nature of video, two video applications contending for the same limited bandwidth can only result in deteriorating the video experience in both sessions. It needs to be ensured that when contention occurs, the user experience for the higher-priority stream is prioritized and that the lower-priority stream is either dropped or forced to accept a lower-bandwidth service. Similarly, there are examples within the same application session. A two-person video conference call may be fine with only two parties, but when the call becomes a three- or four-person session the network may not be able to sustain the required traffic. Applications should negotiate to reserve network resources end to end such as bandwidth, access to low-latency queues, and stable paths before they start the media stream.
Network management tools that provide capacity planning aids, baseline network usage, and deploy and monitor quality of service (QoS) policies can help the network operator in getting to a predictable video application service.
This is the first part of a three-part post. Please read Part 2, where I discuss performance.