unacceptable delay in the voice path. Because
links are used variably, the delay varies with time and may
produce unacceptable jitter in jitter-sensitive applications
such as voice. In the example shown in Figure , the
serialization delay of a 1500-byte packet over a 512-kbps link
will be 24.3 ms. For VoIP traffic, the maximum recommended
one-way, end-to-end delay is 150 ms. Therefore, having a
1500-byte packet ahead of a VoIP packet in the hardware queue
on a 512-kbps link can cause the end-to-end delay of the voice
packet to be over the budget of 24.3 ms. Note
It is
highly recommended that you run the same version of code on
both sides of the WAN link to ensure compatibility and correct
compression results.
Content 4.8 Understanding
WAN Link Efficiency Mechanisms 4.8.6 Link
Fragmentation and Interleaving The use of a hybrid queuing
method such as low latency queuing (LLQ) can provide low
latency and low jitter for VoIP packets while servicing other
data packets in a fair manner. But, even if VoIP packets are
always sent to the front of the software queue, there is still
the issue of serialization delay. A large packet may be on its
way out of the hardware queue, which uses FIFO. When a VoIP
packet is sent to the front of the software queue, the
serialization of the large packet in the hardware transmit
queue can cause the VoIP packet to wait for a long time before
it can be transmitted out. The solution is to fragment the
large packets so that they never cause a VoIP packet to wait
for more than a predefined amount of time. The VoIP packets
must also be allowed to transmit in between the fragments of
the larger packets (interleaving), or there will be no point in
doing the fragmenting. When you are configuring the proper
fragment size to use on a link, a typical goal is to have a
maximum serialization delay of around 10 to 15 ms. Depending on
the LFI mechanisms being configured, the fragment size is
either configured in bytes or in milliseconds, as shown in
Figure .
Content 4.8 Understanding WAN Link
Efficiency Mechanisms 4.8.7 Applying Link
Efficiency Mechanisms Use the following guidelines for
applying link efficiency mechanisms: - Identify slow
links to help determine where the bottlenecks in the network
are located and decide how to apply link efficiency mechanisms
at the appropriate interfaces.
- Calculate the Layer 2
and Layer 3 overhead for each media type that will transport
the business-critical traffic. This process will help you
choose the correct compression type.
- Decide which
type of compression should be used.
- Enable
compression on the WAN interfaces.
Figure summarizes
these guidelines. Example
Figure shows an example of
a network using LFI. Header compression and LFI are typically
configured at the WAN edge for WAN links below T1 or E1 speeds
to optimize the use of the WAN link and to prevent long
serialization delay. Layer 2 payload compression is less
commonly deployed on WAN links, especially without the use of
hardware-assisted payload compression. In this case, use TCP
and RTP compression, as well as LFI mechanisms, because this
network carries converged network traffic.