TABLE OF CONTENTS While we have a protocol structure to support voice, basic IP operates on a best effort philosophy. That means the network provides no guarantees regarding delay, jitter, or loss. A traditional, best effort IP network:
Can lose packets.
Can deliver packets with errors.
Can deliver packets out of sequence.
Has no provision to ensure timing consistency.
Can expose traffic to eavesdropping and other security vulnerabilities.
As we noted, RTP has the ability to resequence voice packets and reestablish timing continuity, but it does not recover lost packets. So the better and more consistent the underlying service, the better the overall voice quality will be. That is where QoS comes in.
The term QoS refers to a variety of techniques used in packet networks, the goal of which is to add predictability to the network service. The characteristics of network performance typically include guaranteed capacity or bandwidth , latency (delay), jitter, and possibly, network availability. To minimize the transit delay, jitter, and packet loss, QoS allows us to recognize voice and video packets and give them a higher priority in the forwarding process.
There are different QoS mechanisms that can operate at either Layer 2 or Layer 3. As the QoS mechanisms span two protocol layers, one of the key design issues in an IP voice network is ensuring that the priority levels are mapped correctly as packets move through the network.
