MPLS Design for VoIP – Protocol Integration
Time sensitive traffic such as voice and video are forwarded
in UDP which operates on the “Send and Pray” transmission philosophy whereby
there is no such mechanism to detect or recover from lost or errored packets.
In short, you’ve got only one shot at delivering a voice packet so a network
that guarantees performance for delay, jitter and loss has a far greater value
in supporting voice and video services.
Building an MPLS based Converged Network
With any guaranteed
packet service such as Frame Relay, there are two critical transmission rates:
the access rate and the service's guaranteed capacity. An access rate
transmission is often linked to the Committed Information Rate (CIR) of any
Link. As long as the average transmission stays below the CIR, the carrier
guarantees to deliver a very high percentage of traffic typically 99.99%. If
the transmission rate on that virtual circuit exceeds the CIR then the
over-lapping traffic is marked as Discard Eligible and will only be delivered
if there is capacity available. These discard eligible packets are usually
discarded when there is congestion condition.
These concepts have been
adopted in MPLS however it has had to be modified in two important ways given
the nature of the MPLS Technology:
1.
Where Frame-Relay essentially offers 2 traffic
categories, guaranteed and discard eligible, an MPLS Network can offer 3 or more;
for convenience those categories are often called Gold, Silver, Bronze and Best
Effort. The Carrier provides different guarantees regarding delay, jitter and
packet loss for traffic sent in each category.
2.
The introduction of CoS (Class of Service)
instead of CIR in contrast to Frame-Relay networks. Such traffic classification
gives an edge on certain percentage of access capacity which is allocated
according to traffic category. The class of service profile is priced proportionally
to the value of service it operates on. Therefore the Higher the class, the
more cost it generates. In this project, such classification is essential and
will have voice traffic as the highest priority assigned. Unless we have other
simultaneous running application, we may not really need such classification in
the project.
How is Voice Traffic Treated in MPLS
Most Network Engineers have not had to deal with the
design of voice networks using MPLS. Taking a closer look at how MPLS actually
works, we do find that there are just 2 main categories:
1. Real
Time
2. Everything
Else
The important difference is how the traffic is treated
between real-time and every other traffic types. It is quite important to
consider the classification of traffic especially if it is treated with high priority.
We all know that IP Voice uses UDP Protocol thus having one chance to transmit
traffic without following up error or corrupted packets along transmission. To
minimize errors or packet loss, different voice encoding modulation systems are
used to compress voice quality thus giving efficient use of capacity and less
packet loss tolerance.
We must make sure that Voice traffic and its
classifications are set with the highest priority due to real-time traffic
class demand. With such configuration (Gold Platform class) we must not
configure too much of voice channels over what the class platform can contain.
Attempting this will only result in packet loss: eg; If you try to configure 10
voice channels over a service with a Gold capacity that can cater for just 5;
you will not have 5 good voice trunks and 5 bad ones, you will have 10 bad
trunks altogether.
Thus, it is rather highly recommended that each class if
well defined accordingly to the traffic types it will transmit and well within
the scope of class capacity that it is transmitted on.
Ref: Michael F. Finneran, "MPLS Design for VoIP - What every user needs to know", July, 1006
