Carrier MPLS support for VoIP

Meeting quality of service guarantees that ensure reliable voice transmissions can be the biggest challenge of deploying VoIP. Implementing Multi-Protocol Label Switching (MPLS) can help enterprises rise to that challenge because the protocol offers network engineers a great deal of flexibility and the ability to send voice and data traffic around link failures, congestion and bottlenecks.

This article discusses how today's network transport carriers provide support for VoIP with their MPLS offerings. It will also explain the best way to leverage a carrier's MPLS services to support the deployment of real-time services.

MPLS and VoIP play well together

MPLS is useful as an enabler for VoIP because it provides Asynchronous Transfer Mode (ATM)-like capabilities with an IP network. Unlike the expensive ATM links that would be required to support VoIP, MPLS provides guaranteed services utilizing IP quality of service on the carrier's backbone. This service and the ability to converge VoIP onto the data network present a tremendous opportunity to reduce operational costs and consolidate infrastructures.

Similarly, VoIP is a major driver for migrating to an MPLS-based environment. In most cases, MPLS is no more cost-effective than other WAN transport options, but it is a more effective solution for transporting VoIP. MPLS may create efficiency regarding the cost of managing a WAN backbone, but in reality, the cost of an MPLS solution is generally more when the cost of supporting real-time traffic is added to the bill.

Real-time services are application services that are susceptible to delay, packet loss and jitter. VoIP and video over IP are considered real-time applications. While other applications such as SAP are vulnerable to delay, VoIP and video over IP (with corresponding voice) are the real focus, because if you cannot deliver these applications with a high degree of confidence that the packets will not be dropped, experience delay or jitter, then you cannot deploy these application services.

How it works

MPLS allows carriers to deliver WAN transport services over an IP backbone using MPLS technology to create virtual circuits, much in the same way that carriers traditionally built ATM and frame circuits over cell-switched backbones. However, there is a major difference; the IP backbones over which the MPLS services are provisioned support Class of Service (CoS) that provides a predictable Quality of Service (QoS) over the IP backbone. This is where support for VoIP is enabled on the carrier's backbone.

Most carriers offer multiple classes of service and some try to differentiate themselves based on the numbers of classes. However, all the carriers offer a class of service that is dedicated to VoIP traffic. This class of service provides the ability for the carrier to service all VoIP traffic prior to servicing any other traffic. The VoIP traffic gets priority over other traffic types (such as email, FTP or batch processing).

Important considerations for VoIP traffic

There are several significant factors that must be understood regarding how to provision and deploy VoIP from a customer perspective. First, there are different types of VoIP traffic. VoIP has two distinct traffic types:

• call signaling traffic (used to set up the VoIP call between two endpoints)
• call bearer traffic (the VoIP packets that make up the actual conversation)

If you think about a WAN link to an MPLS cloud, there is only a certain amount of bandwidth available. With an MPLS offering, it is up to the customer to determine what traffic should go in what CoS bucket and how much bandwidth to allocate to the bucket. The carriers provide rules of thumb, but these generally relate to what traffic to put in what bucket, not how much.

In most cases, CoS 1 is the class into which customers will want to put the VoIP bearer traffic. CoS 2 is good for video and CoS 3 is a good fit for signaling traffic. The carriers usually offer profiles for their CoS offerings with various percentages of bandwidth allocated to each of the classes.

For example, Profile 1 may be as follows:
CoS 1 - 60% of link bw (VoIP traffic)
CoS 2 - 20% of link bw (critical traffic -- SAP eCRM)
CoS 3 - 5% of link bw (signaling traffic for VoIP/video)
CoS 4 - 10% of link bw (email, FTP)
CoS 5 - 5 % of link bw (all other traffic)

Or it could be provisioned like this:
CoS 1 - 40% of link bw (VoIP traffic)
CoS 2 - 30% of link bw (critical traffic -- SAP eCRM)
CoS 3 - 5% of link bw (signaling traffic for VoIP/video)
CoS 4 - 15% of link bw (email, FTP)
CoS 5 - 10% of link bw (all other traffic)

Some carriers offer many, many profiles from which customers can choose. It all depends on their mix of traffic. However, VoIP goes in CoS 1 as it must be delivered before any other traffic.

This is important for VoIP because VoIP represents additional traffic being added to the data link layer. If you utilize a T1 between two sites for data traffic, VoIP traffic will increase the bandwidth significantly, especially since you must reserve bandwidth for VoIP. The carriers offer the profiles; you, as the customer, have to choose which profile maps to the actual traffic flowing over the links. Making this decision can be challenging if you have not yet deployed VoIP, but in reality, the rightsizing of the data links and the choosing of the CoS profiles should be done prior to deploying the MPLS network. This requires a voice traffic study that considers peak call volumes between sites and then the translation of the traditional phone call bandwidth into VoIP bandwidth.

Traditional time division multiplexer voice calls without compression utilize 64Kbs. These can be encapsulated in IP with different codecs. The choice of codec will determine the amount of bandwidth utilized by the VoIP call (plus any L2 or L3 encapsulations). The amount of VoIP bandwidth needed for the peak busy hour is the amount that is needed to be reserved in CoS1. The carrier should have a profile that maps closely to this.

Be warned that carriers support what you buy. Many MPLS carriers use a mechanism called policing on the ingress port of the MPLS router. The carrier will monitor (police) the link and if the reserved bandwidth for CoS 1 is exceeded, the carrier will drop the packets. This will disrupt VoIP traffic. This may seem like a bad thing, but in reality it is a good way of alerting customers when they have oversubscribed their CoS profile and need to increase capacity. This is very critical when sending VoIP over a data network.

MPLS has come a long way and has in truth accelerated the adoption of VoIP. Both VoIP and MPLS took a long time to mature with VoIP maturing sooner. Now that the MPLS offerings are viable and can support VoIP, many organizations are utilizing MPLS transport as an enabler for a true IP-convergent environment.

About the author: Robbie Harrell (CCIE#3873) is the national practice lead for advanced infrastructure solutions at SBC Communications. He has more than 10 years of experience providing strategic, business and technical consulting services. Robbie lives in Atlanta and is a graduate of Clemson University. His background includes positions as a principal architect at International Network Services Inc., Lucent Technologies, Frontway and Callisma.