QoS for IP videoconferencing

Communications News,  June, 2004  

• E-mail
• Print
• Link

Enterprise users are migrating to IP videoconferencing for the same reasons they are adopting IP telephony. They want to leverage a shared IP infrasturcture to reduce travel costs and increase employee productivity and collaboration. Quality of service (QoS) is a major stumbling block to this migration.

Video is sensitive to poor network quality, especially the loss of data packets. In addition, video requires a large amount of bandwidth per call, which can congest enterprise networks. New QoS methods and techniques are required to enable business-quality IP videoconferencing.

The dynamic, connectionless nature of IP networks creates a new set of challenges for IT managers implementing real-time IP videoconferencing. These challenges include the strict latency (end-to-end delay), jitter (variation in delay) and packet loss (packets lost due to congestion) performance requirements that must he met to ensure high quality. Due to the sophisticated compression algorithms used for videoconferencing, a higher level of network performance is required for IP video than for IP voice.

Performance variances can disrupt video performance with freeze frame, checker boxes, streaks and voice-video synchronization problems. Ensuring video-application QoS requires constant monitoring to protect video traffic from congestion and delays resulting from general application traffic. Given the dynamic, "hop-by-hop" nature of router-based IP networks, QoS must he fine-tuned and enforced at every router hop--a complex, resource-intensive task.

Video bandwidth management is another important issue. WAN bandwidth, and especially access bandwidth at the network edge, is a major recurring cost for corporate networks. Because video can consume large amounts of bandwidth, ensuring that an adequate amount of bandwidth is available for videoconferences is critical. Over-provisioning bandwidth for QoS is expensive, while insufficient bandwidth results in poor quality video and user dissatisfaction. Consequently, automated bandwidth management--rather than expensive over-provisioning or excessive manual effort--may be preferable for high-quality video.

IP networks were not initially designed for high-bandwidth, delay-sensitive applications that are impacted by packet loss. Multiple QoS methods dealing with this issue have been developed over the years, including: over-provisioning bandwidth, network monitoring and analysis, differentiated services (a packet classification, marking, policing and prioritization method), and integrated services (per-flow, end-to-end bandwidth reservations).

These techniques, however, cannot guarantee tire videoconferencing quality enterprise users require. Traditional QoS methods can be enhanced to handle the specific challenges of voice and video traffic. These enhancements include:

Video application intelligence. Deep levels of application intelligence and visibility are essential to control and manage video as a session-based service distinct from standard application traffic. Direct, real-time visibility into end-to-end video sessions across the enterprise network is required to ensure bandwidth is continually available throughout the network, packet loss is minimized and delay/jitter are within acceptable limits.

Guaranteed quality of service. Guaranteed QoS is required to replace ISDN-based conferencing and provide enterprise users with a productive, usable service. Video can make use of integrated services' resource reservation protocol (RSVP) to reserve bandwidth across an IP network, providing guaranteed service-quality levels. New aggregated reservation technology solves many of the aforementioned issues of standard RSVP by aggregating multiple video calls end-to-end across the network. This provides the end-to-end reliability of a circuit-based network, with the flexibility and cost of an IP packet network.

Call admission control. Launching new videoconferences when the network cannot support the load degrades performance for all applications. Consequently, call admission control is required to guarantee the quality of existing calls and prevent network congestion. An application-aware, end-to-end call admissions capability, considering both local and network-wide congestion, is required to block calls only when absolutely necessary.

Bandwidth optimization and control. Video is n high-bandwidth application requiring careful network engineering and management. Deep knowledge of the video application and underlying IP network is required to ensure that sufficient bandwidth is available, network wide, for the duration of the videoconference. An automated solution can dynamically adapt to both scheduled and on-demand conferences and network changes, as well as unanticipated traffic spikes.

For more information from Prominence Networks: www.rsleads.com/406cn-260

This article was provided by Richard Dominach, director of business development and marketing at Prominence Networks, Holmdel N.J.