Application Note
Application Note
08 July 2008
Bandwidth
Table of Contents
Revision History:
Revision Date Author Comments
2008-07-08 Scott Beer 1st draft
Bandwidth
1 BANDWIDTH REQUIREMENTS
In a Teleworker, Home Office or Remote Office solution attention to bandwidth
requirements is crucial to a successful deployment. Bandwidth requirements for the
Voice and Video as well as the Data must me understood.
Keep in mind that additional information needs to be added to the Voice Packet to travel
over an IP network, such information as MAC address headers and IP/UDP/RTP
header address information and more. This is about 54 bytes of information, depending
on the network. (There are 8 bits per byte.) With VoIP, the transport medium has
changed since the PCM voice traffic is now carried over an IP network, requiring a
source, destination, User Datagram Protocol (UDP), and Real-time Protocol (RTP)
headers. In the case of G.729, for example, 58 bytes are needed to transport 20 bytes of
compressed Voice. With overhead like this, it is clear that TCP over Internet Protocol
(TCP/IP) is not suited to transporting compressed voice (or very small packets). An
alternative protocol is to use RTP header extensions in combination with UDP over IP.
In addition, the PCM 'stream' needs to be 'packetized', so now there is a Sample Rate.
Taking a continuous 64kbps Stream and sampling it down to 20 ms of 'sampled' audio.
Sending now 50 packets per second to ensure the 64kbps of voice data. Keep in mind,
the more frequent the sample rate the more network overhead is added, to the point
where there will be significantly more overhead then data.
Compression Algorithms:
Standardized compression algorithms have since been developed for use in voice
applications. Data Compression is the process of encoding information using fewer bits
than an unencoded representation would use through use of specific encoding schemes.
The most popular compression algorithms are G.723 (6 Kbps every 30 ms), G.729 (8
Kbps every 20 ms), and uncompressed G.711 (64 Kbps every 20 ms). The main
differences between them are packet size, voice quality, and complexity.
Bandwidth pg. 1
1.1.1 Calculating VoIP Bandwidth
The bandwidth allocation is dependent on the codec or compression type used for the
VoIP encoders. For VoIP using G.723 compression, a minimum bandwidth of at least
32 Kbps is required, while G.729 compression requires between 30-40 Kbps of
bandwidth depending on the sampling rate (20 or 30 ms).
Here is a practical application of these formulas, assuming the use of the G.711
CODEC:
Voice packet size (214 bytes) = MAC header (14 bytes) + IP/UDP/RTP header
(40 bytes) + voice payload (160 bytes)
Voice packet size in bits (1712) = voice packet size (214) * bits per byte (8)
Voice packets per second (50) = CODEC bit rate (64 Kbps) / voice payload size
(1280 bits)
Bandwidth per call (85.6 Kbps) = voice packet size in bits (1712) * voice packets
per second (50)
VoIP‟s impact on the network can be estimated by taking the bandwidth per call and
extrapolating with the number of concurrent calls that are expected to be handled during
peak busy-hour calling times, as previously calculated. Such estimates are usually made
with the largest amount of bandwidth required by a CODEC, which is currently G.711.
Once the call flows and call volumes have been identified and the available network
bandwidth has been determined, the next step is determining how much bandwidth will
be required to support the expected VoIP traffic.
Keep in mind the network connectivity, whether Full Duplex, Half Duplex or
Synchronous or Asynchronous. As this has bearing on how much actual throughput is
available. As Half Duplex and Asynchronous network connectivity do not have the
same Bandwidth in both directions.
Bandwidth pg. 2
Additional VoIP Bandwidth Requirements Using VPN
IPsec is one common method for encryption. IPsec (IP security) is a standard for
securing Internet Protocol (IP) communications by encrypting and/or authenticating all
IP packets. IPsec provides security at the network layer. It is made of a set of
cryptographic protocols for securing packet flows and key exchange. Cryptographic
protocols such as Encapsulating Security Payload (ESP), which provides authentication,
data confidentiality and message integrity; and Authentication Header (AH), which
provides authentication and message integrity, but does not offer confidentiality. These
additional cryptographic protocols add approximately 10 per cent overhead to the voice
traffic over a VPN tunnel.
Keep in mind the network connectivity, whether Full Duplex, Half Duplex or
Synchronous or Asynchronous. As this has bearing on how much actual throughput is
available. As Half Duplex and Asynchronous network connectivity do not have the
same Bandwidth in both directions.
There are many video multimedia compression formats, too many to discuss the
bandwidth requirements of each in this document. Video Bandwidth requirements are
dependent on a dynamic assignment of Resolution with the selected Compression
Algorithm and by Frame sizes which all combine to create a „Black Art” of
understanding of Video compression and bandwidth requirements.
Streaming video quality is dependent in part upon the video encoding process and the
amount of bandwidth required for it to be viewed properly. Encoding a video so that it
will stream at a speed for a user who has limited or low bandwidth, requires a high
degree of compression is applied to both the video and audio tracks. This process
eliminates portions of the audio and video data, which is why most videos you see don't
appear or sound as clear as you'd like.
Bandwidth pg. 3
A user who is connected to the Internet using a high speed cable modem or DSL
connection, for example, can watch any streaming video clip which has been encoded for
transmission at their connection speed or lower. However, a user who is connected to
the Internet using a 56k modem who attempts to view a streaming video clip which has
been specifically digitized and encoded for transmission to cable modem users, for
example, will get very choppy video which plays for a second or two and then pauses for
several seconds until more video data is transferred to their PC.
Here is a small sample of the bandwidth possibilities using H.264 video compression.
Max video
Max video
bit rate
bit rate Max video Examples for high
Max video (VCL) for
Max Max frame (VCL) for bit rate resolution @
Level bit rate High 4:2:2
macroblocks size Baseline, (VCL) for frame rate
number (VCL) for and
per second (macroblocks) Extended High 10 (max stored frames)
High Profile High 4:4:4
and Main Profile in Level
Predictive
Profiles
Profiles
[email protected] (8)
1 1485 99 64 kbit/s 80 kbit/s 192 kbit/s 256 kbit/s
[email protected] (4)
[email protected] (8)
1b 1485 99 128 kbit/s 160 kbit/s 384 kbit/s 512 kbit/s
[email protected] (4)
[email protected] (9)
1.1 3000 396 192 kbit/s 240 kbit/s 576 kbit/s 768 kbit/s [email protected] (3)
[email protected] (2)
[email protected] (7)
1.2 6000 396 384 kbit/s 480 kbit/s 1152 kbit/s 1536 kbit/s
[email protected] (6)
[email protected] (7)
1.3 11880 396 768 kbit/s 960 kbit/s 2304 kbit/s 3072 kbit/s
[email protected] (6)
[email protected] (7)
2 11880 396 2 Mbit/s 2.5 Mbit/s 6 Mbit/s 8 Mbit/s
[email protected] (6)
[email protected] (7)
2.1 19800 792 4 Mbit/s 5 Mbit/s 12 Mbit/s 16 Mbit/s
[email protected] (6)
[email protected](10)
2.2 20250 1620 4 Mbit/s 5 Mbit/s 12 Mbit/s 16 Mbit/s
[email protected] (7)
[email protected] (6)
Bandwidth pg. 4
Max video
Max video
bit rate
bit rate Max video Examples for high
Max video (VCL) for
Max Max frame (VCL) for bit rate resolution @
Level bit rate High 4:2:2
macroblocks size Baseline, (VCL) for frame rate
number (VCL) for and
per second (macroblocks) Extended High 10 (max stored frames)
High Profile High 4:4:4
and Main Profile in Level
Predictive
Profiles
Profiles
[email protected] (12)
[email protected] (10)
3 40500 1620 10 Mbit/s 12.5 Mbit/s 30 Mbit/s 40 Mbit/s
[email protected] (6)
[email protected] (5)
[email protected] (13)
3.1 108000 3600 14 Mbit/s 17.5 Mbit/s 42 Mbit/s 56 Mbit/s [email protected] (11)
[email protected] (5)
[email protected] (5)
3.2 216000 5120 20 Mbit/s 25 Mbit/s 60 Mbit/s 80 Mbit/s
[email protected] (4)
[email protected] (9)
4 245760 8192 20 Mbit/s 25 Mbit/s 60 Mbit/s 80 Mbit/s [email protected] (4)
[email protected] (4)
[email protected] (9)
4.1 245760 8192 50 Mbit/s 50 Mbit/s 150 Mbit/s 200 Mbit/s [email protected] (4)
[email protected] (4)
[email protected] (4)
4.2 522240 8704 50 Mbit/s 50 Mbit/s 150 Mbit/s 200 Mbit/s
[email protected] (4)
[email protected] (13)
[email protected] (13)
5 589824 22080 135 Mbit/s 168.75 Mbit/s 405 Mbit/s 540 Mbit/s [email protected] (12)
[email protected] (5)
3680x1536/26.7 (5)
[email protected] (16)
5.1 983040 36864 240 Mbit/s 300 Mbit/s 720 Mbit/s 960 Mbit/s [email protected] (5)
[email protected] (5)
Bandwidth pg. 5
2 VoIP Performance
VoIP has three specific performance requirements that have to be met in order to
provide a quality voice conversation.
Latency
Anyone who has tried a long distance conversation over a satellite link may have
experienced how excessive latency impacts voice quality. These long delays make it
difficult for callers to have a natural speech patter to their conversation as it is difficult to
determine when the person at the other end has finished talking. How much latency is
too much? A rule of thumb is that one-way latency should not exceed 150 milliseconds.
150 millisecond delays are noticeable but tolerable, but when latency exceeds 250
milliseconds it becomes too difficult to conduct a conversation. Latency is a non-issue
on the PSTN, but delays on IP networks can easily cause latency to exceed 150
milliseconds due to data and voice congestion.
Even when voice packets are not blocked by data packets they are subject to their own
serialization delay – the amount of time that is takes to clock the bits onto a serial link.
Again, this delay is determined by packet size and link speed. Reductions in packet size
result in less serialization delay and therefore, lower end-to-end latency. The adjustment
in the sample rate will adjust the packet size, but the more packets sent onto the network
increases the overhead added to each conversation.
Jitter
Another key performance metric is jitter. Jitter is the amount of variation in latency that
is experienced over time. SIP phones have some ability to buffer incoming audio
streams to compensate for jitter, but excessive jitter can disrupt conversations. A rule of
thumb is that Jitter should not exceed 50 milliseconds. Again, the PSTN has virtually no
latency and therefore no jitter, but enterprise IP networks are subject to jitter caused by
congestion on LANs and WANs and by packet buffering in routers and other network
devices.
Bandwidth pg. 6
To avoid Jitter, ensure reliable network connectivity. Planning and deploying a network
with no points of congestion and adequate bandwidth throughout the network.
Networks deployed as „Flat‟ as possible with enough bandwidth to ensure proper data
and voice communications result in the best performance. Avoid chaining multiple
Layer 2 switches together in a long serial chain, as the higher the up the chain traffic goes
the more congested the final up link becomes. Deploy and number of Quality of Service
techniques to ensure the best prioritization of Voice and Data. Calculate the bandwidth
requirements of the voice and data at all common points of convergence.
Packet Loss
The third metric is packet loss. Since VoIP is a real-time audio service that uses UDP
transport protocols, there is no way to recover lost packets. Packet loss can result in a
metallic sound or dropouts in conversations that can be very frustrating to users. The
PSTN experiences virtually no loss of digitized voice, but IP networks routinely
experience packet loss for a variety of reasons but primarily due to congestion.
To avoid Packet Loss, ensure reliable network connectivity. Planning and deploying a
network with no points of congestion and adequate bandwidth throughout the network.
Networks deployed as „Flat‟ as possible with enough bandwidth to ensure proper data
and voice communications result in the best performance. Avoid chaining multiple
Layer 2 switches together in a long serial chain, as the higher the up the chain traffic goes
the more congested the final up link becomes. Deploy and number of Quality of Service
techniques to ensure the best prioritization of Voice and Data. Calculate the bandwidth
requirements of the voice and data at all common points of convergence.
Bandwidth pg. 7
3 Quality of Service (QoS)
DiffServ operates on the principle of traffic classification, where each data packet is
placed into a limited number of traffic classes, rather than differentiating network traffic
based on the requirements of an individual flow. Each router on the network is
configured to differentiate traffic based on its class. Each traffic class can be managed
differently, ensuring preferential treatment for higher-priority traffic on the network.
The DiffServ model does not make judgment on what types of traffic should be given
priority treatment since that is left up to the network operator. DiffServ simply provides
a framework to allow classification and differentiated treatment. DiffServ does
recommend a standardized set of traffic classes (discussed below) to make
interoperability between different networks and different vendors' equipment simpler.
DiffServ does not extend onto the Internet (WAN). This mechanism is meant for local
networks (LAN) only.
Bandwidth pg. 8
3.1.2 WAN QoS Options
There are no protocols or standards that can be reliably provide Quality of Service over
the Internet (WAN). The Internet is intended as a “Best Effort” network without
concerns for reliability and latency.
There are some Internet Service Providers that will offer some Quality of Service
capability within their own network. Some use MPLS service, and others may continue
to use DiffServ to provide QoS within their own network.
Traffic shaping provides a mechanism to control the volume of traffic being sent into a
network (bandwidth throttling), and the rate at which the traffic is being sent (rate
limiting). For this reason, traffic shaping schemes are commonly implemented at the
network edges to control traffic entering the network. This control can be accomplished
in many ways and for many reasons but traffic shaping always simply consists in delaying
packets. Traffic policing is the related practice of packet dropping and packet marking.
Traffic shaping can be applied by the traffic source (for example, computer or network
card) or by an element in the network.
Ingate has incorporated a Quality of Service Module within our products. With this
application the SMB/SME can shape traffic to better utilize their Internet bandwidth.
They can prioritize and do bandwidth limitation for different traffic types, based on
source or destination IP addresses, type of service, packet size, and TOS/DSCP markup.
This ensures the VoIP traffic has primary use of the available bandwidth.
Although traffic shaper technology can control the data in and out of a SMB/SME at the
LAN-WAN edge, it does not control or provide a Quality of Service over the Internet.
Bandwidth pg. 9