VoIP: Are the Networks Ready for Prime Time?

By Marc Coluccio

Enterprises of all sizes are quickly adopting IP telephony systems. Cost reduction, improved productivity, and better collaboration are some of the measurable benefits of IP solutions. However, ensuring high quality voice conversations in your call center requires more than just buying the latest equipment.

Successfully deploying Voice over IP (VoIP) requires a deeper understanding of the nuances of delivering application priority and jitter control over your company’s network infrastructure. Most legacy network deployments and Internet based VPN (Virtual Private Network) services were not created with VoIP requirements in mind. Therefore, making sure your network is ready for IP telephony is a critical factor in determining how functional the service will be and how well it will be received by end users.

The fundamental requirement to achieve toll-quality voice is to deploy an IP PBX or IP Centrex Service, accompanied by IP handsets or soft-phones, over a properly designed network infrastructure. The LAN/WAN infrastructure must deliver sufficient throughput and meet latency, jitter, and packet loss requirements. For best results, it also should support application differentiation and provide failover and load-balancing options.

Delivering sufficient throughput: The amount of bandwidth required for voice depends on the number of simultaneous calls, the voice encoding scheme used in the IP handset or softphone, and the signaling overhead. The two most commonly used codecs for VoIP deployments include: International Telecommunications Union (ITU) G.711 and G.729. The bandwidth requirements for each codec are shown in Fig. 1.

Meeting latency and jitter requirements: Latency is the time it takes for a data packet to go from one location to another. Variables that can cause excessive latency include: physical distance, number of intermediary routers, encryption and decryption, and converting your voice into a data stream and then back to voice on the far end. If latency is too high, it can interrupt the natural flow of a conversation by causing the two parties to confuse latency for pauses in each other’s speech. VoIP latency requirements are shown in Fig. 1.

Jitter is the variation in latency over the LAN and WAN; it is caused by IP telephony packets arriving in uneven patterns at their destination. Jitter has many sources: network congestion, queuing methods used in routers and switches, or routing options such as MPLS or frame relay used by carriers. Users hear jitter as degraded voice quality and excessive jitter can render a VoIP conversation quite unusable.

Packet loss requirements: Packet loss results in a metallic sound or dropouts in the conversation. Packet loss is typically caused by network congestion, which often results from incorrect implementation of application priority.

CM June 2005 #10

If you are deploying IP Telephony across a WAN built on Frame Relay, VPN, or Internet circuits, the lack of latency and jitter control, combined with the inability to prioritize packets through the network core, often results in intermittent call quality and completion issues.

Common issues with Internet and Frame Relay include:

  • Buying more bandwidth increases costs yet fails to permanently address application performance problem or voice quality. You must add intelligence and application awareness to your network to permanently fix the problem. Most WAN services do not support this functionality including frame relay and VPN solutions.
  • Buying the best LAN routers and equipment isn’t a solution; your WAN provider must support application differentiation and priority queuing throughout the network core.

A well thought out IP Telephony deployment will be built on a LAN/WAN that has strict control over latency, throughput, and jitter parameters. Application differentiation and prioritization should be seamless from the LAN to the WAN and configurable to your exact needs. Some Multi-Protocol Label Switching (MPLS) solutions can achieve latency, throughput, and jitter control but often fall short in matching priority settings on a per customer basis.

Virtually routed architectures can provide latency, throughput, and jitter guarantees and can also customize application differentiation and priority on a per customer and per location basis. This allows end-to-end packet prioritization between remote sites and ensures toll-quality voice calls. With a Virtually routed solution, you’ll be able to:

  • Determine which applications, users, and locations consume the most bandwidth.
  • Ensure network priority for critical applications, such as VoIP, SAP, Oracle, Citrix, and email.
  • Block and contain undesirable frivolous and malicious traffic (both downstream and upstream).
  • Enjoy 99.999% network availability through novel failover and load-balancing options.

Marc Coluccio is certified by Microsoft, Cisco and CompTIA; he has implemented frame-relay and virtual private wide-area networks.

[From Connection Magazine June 2005]