BACKGROUND IMAGE: iSTOCK/GETTY IMAGES
Voice over IP (VoIP) technology has evolved to the extent where businesses can safely rely on it, yet enterprises...
must understand there are certain caveats that need to be considered to ensure a VoIP deployment is a successful one.
Real-time based data such as VoIP traffic and voice applications tolerate minimal variation in the amount of delay affecting the delivery of their voice packets. VoIP traffic does not tolerate packet loss or jitter, as they can dramatically degrade the quality of the voice transmission delivered to the end user.
To effectively transport voice traffic over IP, it is necessary to implement mechanisms that ensure VoIP traffic is given the proper priority along its path. The Internet Engineering Task Force (IETF) has defined various quality of service (QoS) models that aid the implementation of QoS in an IP network, from one end to the other.
Three main models are used for implementing QoS within a network infrastructure: Best-Effort, Integrated Services (IntServ) and Differentiated Services (DiffServ).
The Best-Effort model is the model upon which the Internet works. This model applies to business networks where no QoS policies have been configured or if their infrastructure does not support QoS.
With the Best-Effort model, there is no guaranteed delivery of packets, and all packets receive the same treatment and priority. For example packets that are part of a VoIP stream are treated with the same priority with packets that are part of an email session that might be pure spam.
Unfortunately, many businesses that have implemented VoIP still use the Best-Effort model, as the network infrastructure either does not support QoS or has not been properly configured.
The IntServ model aims to reserve bandwidth along a specific path in the network, guaranteeing from one end to the other the necessary bandwidth for mission-critical applications such as VoIP.
IntServ follows the signaled-QoS model where the end-hosts signal their QoS requirements to the network. Every individual communication stream needs to request resources from the network. Edge routers use the Resource Reservation Protocol (RSVP) to signal and reserve the desired bandwidth for each flow in the network.
One major drawback with the IntServ model is every device along the path of a packet must be fully aware of RSVP and have the ability to process the signaling of the required QoS. This includes routers, servers, PCs and other equipment.
Another drawback is that reservations in each device along the path need to be periodically refreshed, adding additional traffic and overhead along the network. These types of reservations are called "soft-states" and maintaining them in each router along with the admission control at every hop means increased memory and CPU requirements for a large number of reservations. This adds unnecessary complexity to the network infrastructure, making it difficult to maintain.
The DiffServ QoS model for enterprise networks
The DiffServ model is the most popular QoS model used throughout enterprise networks. It was designed to overcome the limitations of Best-Effort and IntServ models, while maintaining the ability to provide an almost guaranteed QoS.
DiffServ works on the provisioned QoS model where network elements, such as routers and switches, are configured to service multiple classes of traffic with different priorities. QoS characteristics -- such as bandwidth, delay and prioritization -- are configured on a hop-to-hop basis along the network infrastructure, making DiffServ cost-effective and scalable.
For the DiffServ QoS model to work, network traffic must be divided into classes that are based on the company’s requirements. From there, each class is assigned a different level of service using various policies. Network devices identify traffic as it passes through them and enforce the configured policies, making sure that each class/service is served as instructed.
Voice traffic from IP phones is almost always given prioritized treatment over other types of network traffic. Even though DiffServ cannot guarantee the service quality it provides, when configured carefully, it can perform wonders.
The DiffServ QoS model makes use of three terms to help describe its functionality:
DSCP: Marking of a packet, used to select a QoS treatment for a packet. DiffServ uses 6 bits of the original Type-of-Service (ToS) field (analyzed below) within the IP Header, to mark packets.
Behavior Aggregate (BA): Collection of packets with the same DSCP value, crossing a particular direction within the network.
Per-Hop-Behavior (PHB): PHB refers to the QoS techniques applied to a packet. This includes a variety of techniques such as queuing, prioritization, traffic shaping, expedited forwarding (EF), assured forwarding (AF) and others.
To divide traffic into classes and mark traffic, DiffServ makes use of the ToS field within the IP header.
Below is a representation of the original ToS field located within the IP Header, as defined by Requests for Comments RFC 1122 and RFC 1349:
The original ToS field has now been replaced and used by the Differentiated Services QoS model as shown below and defined by RFC 2474:
As shown, the DSCP field, used by the DiffServ model to mark packets, consists of 6 bits, out of the 8 available bits.
VoIP technology relies heavily on proper QoS mechanisms throughout the network infrastructure. Businesses invest thousands of dollars in new technologies such as VoIP, but usually fail to give the same priority in setting up QoS within their network boundaries. As a result, many businesses suffer from problems that derive from poor or non-existent QoS implementation, creating a negative impact to their VoIP experience.
Get started with the SearchUnifiedCommunications' VoIP quality of service and performance resource page
Telephony expert Jon Arnold explains the importance of quality of service when measuring the performance of voice service.