OPTIMIZING PERFORMANCE FOR VOICE OVER IP AND UDP TRAFFIC

Voice-over-IP (VoIP) is a special category of network traffic that requires separate and distinct consideration from the more traditional TCP-based traffic found in IP networks. Certainly, user expectations about the quality and availability of telephony services are at a much higher level than for web and file applications. Due to these elevated user expectations, VoIP traffic must be handled with care. Improper configuration and support of VoIP traffic can lead to degraded voice quality and dropped calls. Some WAN optimization vendors have been making noteworthy claims regarding their ability to “accelerate” and “improve” VoIP traffic. The vendors suggest that they have exclusively discovered the technology to deliver significant compression results for VoIP traffic while also improving voice quality. But just how credible and reliable are these claims and statements? While they certainly sound interesting, these claims are simply not supported by the facts. VoIP has Special Challenges VoIP is considered real-time traffic; VoIP datagrams must be delivered by the network with a minimal amount of jitter in order to maintain voice quality. Latency must be kept below 150ms if possible. Any deviations from these standards will affect voice quality. For these reasons, VoIP traffic must receive premium network service through any IP network, and any techniques employed to expedite delivery of VoIP traffic must not add latency or jitter. To “accelerate” VoIP traffic is an ambiguous, often-misleading phrase. VoIP traffic moves at the speed of light, just like all other types of IP traffic. It can be delayed by network congestion and various other issues, resulting in degraded voice quality and dropped calls. But to “accelerate” VoIP traffic does not make sense in the context of the network (you cannot make VoIP packets travel faster than the speed of light) or the telephony application (you cannot make the participants talk any faster). The remainder of this document will explore the various means used to expedite and optimize delivery of VoIP traffic across WANs. Various product vendors offer each of these techniques – some approaches are worthwhile, while others are of questionable value at best. Compressing VoIP Traffic Each VoIP conversation does not represent a significant amount of traffic – a few kbps or so. While many vendors may draw attention to their “impressive” compression results for VoIP traffic, it is important to remember that in most WANs, the baseline byte-volume of VoIP traffic pales in comparison to the huge volume of other TCP/IP traffic. Even if significant compression results could be achieved, there will not be a significant amount of bandwidth saved when considering the total volume of all traffic in the IP network. VoIP uses IP datagrams to transport analog data that has been encoded into binary format through efficient algorithms. These algorithms have been developed after many years of engineering effort, and the resulting encoded data generated from a normal phone conversation has little or no repetition and entropy. As a result, there are very few opportunities to compress the VoIP datagram payload without affecting voice quality. Instead, the primary opportunities to compress VoIP traffic come from a technique known as header compression. Header compression attempts to leverage the repetitive byte values that are found in the VoIP packet header. One distinguishing characteristic of VoIP that makes this somewhat worthwhile is the significant size of the header relative to the size of the entire datagram, especially when compared to “normal” TCP/IP traffic. However, the opportunity is limited because the size of the header is rarely larger than the size of the VoIP payload, and claims to be able to deliver greater than 30% compression through header compression techniques simply do not add up. In addition, header compression technology is integrated into most WAN routers. If the limited gains from header compression are worth pursuing, then a separate WAN optimizer product is not necessary. Furthermore, header compression techniques that are based on tunneling and coalescing of packets will add jitter and latency to the end-to-end VoIP traffic. That is because some packets are delayed so that they can be buffered up and coalesced with other packets through use of common tunneling headers. While these techniques can effectively reduce and eliminate some VoIP headers, the added jitter and delay will affect the quality of the voice conversation. The greater the aggressiveness with which these compression techniques are applied, the greater the resulting jitter and delay, and negative impact on overall voice quality. As a result, many network administrators simply choose not to apply header compression techniques due to their impact on the voice application.

Next