Traffic shaping (also known as bandwidth shaping or packet shaping) is a method of regulating network data flows to ensure a consistent throughput or quality of service (QoS) for specific applications. It involves delaying the transfer of low-priority packets while expediting the transfer of high-priority packets to achieve a desired traffic profile. Restricting the flow of incoming packets is called bandwidth throttling or bandwidth policing, and restricting the flow of outbound packets is called rate limiting.
Traffic shaping is useful when real-time data (such as Voice over IP traffic) needs to be given priority over data that is less time-sensitive (such as email or web browser traffic). Bandwidth is an expensive resource that must be shared among many applications. Without traffic shaping, high-volume non-time-sensitive traffic would overload your network interfaces, resulting in delay, jitter and packet loss for real-time voice streams. Traffic shaping uses the concepts of data classification, policy rules, QoS and queuing to guarantee sufficient network bandwidth for voice and other time-sensitive mission-critical applications.
Traffic shapers use advanced queuing technology to improve upon the standard first-in-first-out (FIFO) approach to queuing. As packets arrive at the packet shaper, they are stored in a buffer for processing. High-priority packets are retransmitted immediately, while less important packets are delayed before transmission. Buffers have a finite capacity, and must handle situations where the buffer is full. One common solution is to simply drop arriving packets (known as tail drop), which effectively results in traffic policing as well as shaping.
Shaping is important for voice due to the real time nature of the traffic. When you listen to a conversation, you need to hear and mentally process the information almost instantaneously. If parts of the conversation go missing, you won’t understand what is being said. If voice packets arrive too late, even by a second or two, they must be discarded because they can’t simply be inserted into a later part of the conversation.
This is quite different to most other network applications. If several packets from an email message or file transfer arrive late, they are simply merged back into the final file at the destination. A few seconds of delay will cause no adverse effects.
That’s why it’s essential to reserve some bandwidth just for voice, and to shape the voice traffic to ensure it arrives in a constant, steady, reliable stream at the destination. It doesn’t matter if parts of an email arrive out of sequence, because those parts will have been recompiled in the correct order before you read the message. But for a phone call you must hear everything arrive in the correct order, in real time.
Traffic shaping should be configured at any location where your employees will use IP phones, including your head office, satellite offices, and home offices. And because voice packets are very small, you typically only need to dedicate a small amount of bandwidth for voice, so there will be little or no impact on the performance of your other network applications.