Cisco Cisco Customer Response Solution Downloads Guía De Diseño

Descargar
Página de 93
 
6-4
Cisco Unified Contact Center Express Solution Reference Network Design, Release 4.1
Chapter 6      Bandwidth, Security, and QoS Considerations
  Estimating Bandwidth Consumption
If the VoIP provider is a VoIP Monitor service, the requestor is a Recording service, and these 
services reside on the same machine, there is no additional bandwidth used on the network to record 
the call.
Regardless of who the requestor and VoIP provider are, the bandwidth requirement between these two 
points is the bandwidth of the IP call being monitored and/or recorded. You can think of each monitoring 
and/or recording session as being a new phone call (2 voice streams) for calculating bandwidth. 
Therefore, to calculate bandwidth to support the Silent Monitoring feature, you can use the same 
calculations used to provision the network to handle call traffic.
IP Call Bandwidth Usage
An IP phone call consists of two streams of data. One stream is sent from phone A to phone B. The other 
stream is sent from phone B to phone A. The voice data is encapsulated into packets that are sent over 
the network. The amount of data required to store a voice stream is dependent upon the CODEC used to 
encode the data. The CAD software can support both the G.711 and G.729 CODEC.
The voice data itself is transmitted over the network using the Real-Time Transport Protocol (RTP). The 
RTP protocol supports the idea of silence suppression. When silence suppression is used, no voice 
packets are sent over the network if there is not sound. Otherwise, even packets that contain silence are 
sent. This lowers the average required bandwidth for a call. Although CAD supports silence suppression, 
the lower bandwidth requirements for silence suppression should not be used when provisioning the 
network because the worst case scenario would be where there is not silence in the call, requiring the 
full call bandwidth as if silence suppression was not enabled.
When calculating bandwidth for an IP call, you must use the size of the RTP packet plus the additional 
overhead of the networking protocols used to transport the RTP data through the network. 
For example, G.711 packets carrying 20 ms of speech data require 64 kbps (kilobytes per second) of 
network bandwidth per stream. These packets are encapsulated by four layers of networking protocols 
(RTP, UDP, IP, and Ethernet). Each of these protocols adds its own header information to the G.711 data. 
As a result, the G.711 data, once packed into an Ethernet frame, requires 87.2 kbps of bandwidth per 
data stream as it travels over the network. Since an IP phone call consists of two voice streams, in this 
example, a call would require 174.4 kbps.
The amount of voice data in a single packet also influences the size of the packet and bandwidth. The 
example above used packets containing 20 milliseconds of speech for its calculations, but this value can 
be changed in the Cisco Unified CallManager configuration for each supported CODEC. Configuring 
packets to contain more speech information reduces the number of packets sent over the network and 
reduces the bandwidth since there are fewer packets containing the additional networking headers, but 
the packet sizes increase. 
 shows the bandwidth required for a phone call for the different 
combinations of CODEC and amount of speech per packet.
Table 6-1
Per-Call Packet Size Bandwidth Requirements
CODEC
Milliseconds of speech per 
packet
Bandwidth required (Kbps) for a 
call 
G.711
10
220.8
G.711
20
174.4
G.711
30
159.0
G.729
10
108.8
G.729
20
62.4
G.729
30
47.0