Page 122 of 378 ITG Engineering Guidelines
553-3001-202 Standard 1.00 April 2000
Routin
g
and hop count
Each site pair takes different routes over the intranet. The route taken
determines the number and type of delay compone nts that add to end-to-end
delay. Sound routing in the network depends on correct network design at
many levels, such as the architecture, topology, routing configuration, link
and speed.
ITG s
y
stem dela
y
The transmitting and receiving ITG nodes together contribute a processing
delay of about 33 ms to end-to-e nd delay. T hi s is t he amoun t of t ime requ ired
for the encoder to analyze and packetize speech, and by the decoder to
reconstruct and depacketize the voice packets.
There is a second component of delay which occurs on the receiving ITG
node. For every call terminating on the receiver there is a jitter buffer which
serves as a holding queue for voice packets arriving at the destination ITG.
The purpose of the jitter buffer is to smooth out the effects of delay variat ion
so that a steady stream of voice packets can be reproduced at the destina tion.
The default jitter buffer delay for voice is 60 ms.
Other dela
y
components
There are other delay components but they are generally considered very
minor.
• Ro uter processing delay. The time it takes to forward a packet from one
link to another on the router is the transit or ro uter proc essing del ay. In a
healthy network, router processing delay is on the order of a few
milliseconds.
• LAN segment delay. The transmission and processing delay of packets
through a healthy LAN subnet is on the order of just one or two
milliseconds.
Reduce link delay
In this and the next few sections, the guidelines examine different ways of
cutting down one-way delay and packet loss in the ITG network.
The time it takes for a voice packet to be queued on the transmission buff er
of a link until it is received at t he n ext hop ro ute r is t he l ink del ay. Link d elay
can be reduced by: