Cascading MCUs
As we can see, the limit of 15 conference users per MCU is not adequate for
many larger conferences.To overcome this limitation, we can use a feature known
as cascading. Cascading allows two or more conferences managed by separate
MCUs to be joined together, in order to produce a much larger conference.
Participants of each conference, meanwhile, will be unaware of the cascaded
nature of the conference.
Cascading MCUs is also an excellent way of distributing processing load, and
enabling the local processing of voice/video data streams.When a user joins a
cascaded conference, she attaches to the nearest MCU. Each individual participant
using this particular MCU will have their own voice/video stream to the MCU.
However, if there is another remote MCU with another group of participants
that is part of the same conference, only a single voice/video data stream flows
between the MCUs.A MCU carries out processing functions for only the participants
attached to it.
The number of MCUs that can be cascaded into the same conference
depends on available bandwidth. By placing MCUs on different segments, it is
possible to reduce network congestion caused by voice/video traffic, and to
increase the potential number of participants in a conference.
Once a conference is established on a MCU, other MCUs are added by invitation.
Once a MCU is invited, and successfully joins a conference, participants
using this MCU can invite further MCUs to join. Figure 4.7 illustrates a conference
between two cascaded MCUs.
www.syngress.com
118 Chapter 4 • AVVID Clustering
Figure 4.7 shows how bandwidth consumption can be reduced by distributing
MCUs across different network segments. In this example, seven participants
are connected to the MCU on LAN A at 384 Kbps, with four on LAN B
at 384 Kbps.The MCUs are cascaded together, totaling eleven participants at 384
Kbps.This would not be possible if a single MCU were used, for it would only
allow a maximum of nine users at 384 Kbps.The options available would be
either to reduce the video quality, or, as in this case, to cascade two MCUs.
However, the major advantage of this solution is that by using two MCUs we are
reducing the communications across the wide area link to a single stream. If a
single MCU at LAN B were used, the seven participants on LAN A would have
to register with it.This would mean 7 x 384 Kbps = 2,688 Kbps of traffic
traversing the wide area link.