Specifications

MBG Engineering Guidelines, Release 8.0
Now a two-way call is properly routable between the two servers.
Note: This feature is suitable only for small numbers of servers. For N servers, each server requires a list of N-1
translation rules. This becomes difficult to manage for larger values of N. An auto-population feature, leveraging
the clustering support, is being investigated for a future release.
11.3 Streaming Addresses
The MBG server will automatically determine the correct IP addresses to which endpoints must send their
(S)RTP, if the server has been put into a standard, supported configuration and the correct network profile for
that configuration has been chosen. See Services for full details on the supported configurations.
However, sometimes it is necessary to override the default streaming addresses, typically due to a non-standard
configuration. When the admin views the Network profiles (under the Configuration tab), the current network
profile will be shown with an interface to apply the supported network profiles.
Arbitrary addresses can be entered by selecting the Custom profile, if the “canned” configurations are not
suitable. These addresses are used during signaling to inform the endpoints where to send RTP. If they are
incorrect, there will be audio problems (typically one-way audio or no audio).
DMZ Deployment Profile
When the MBG server is deployed in a DMZ (see Figure 3 on page 5), all endpoints should send RTP to the IP
address seen by the Mitel AMC during a sync operation. This is the MBG server's publicly-visible address after
any NAT by the DMZ firewall. This address is configured automatically when DMZ mode is selected.
If this address is incorrect for some reason (such as multiple layers of NAT), use the Custom profile to enter the
correct address.
Gateway Deployment Profile
A standard example of the gateway deployment is shown in Figure 2 on page 4. In this configuration, the WAN
interface on the server is routable on the “public” network. This is referred to as the “set-side” streaming address,
to which teleworker devices and service provider trunk equipment sends RTP. The “public” network is typically
the Internet, but may instead be a managed service-provider network, MPLS LAN extension, corporate network,
etc.
The LAN interface is attached to the private network. This is referred to as the “ICP-side” streaming address, to
which LAN devices (phones), conference bridges, ICPs, voicemails, etc send RTP. The private and public
addresses must be on different IP networks.
The defaults are usually acceptable. However, if not, they can be changed by using the Custom profile.
11.4 RTP Frame Size
By default, the RTP frame size is auto-negotiated between the endpoints. A size of 20 ms is common in North
America, with 40 ms becoming common in Europe. If needed, the administrator can force use of a particular
frame size. For example, some SIP trunk service providers insist on a particular RTP frame size.
The Configuration tab holds the global master setting. This setting is used as the default, and should be left on
“Automatic” unless there is a pressing need to change it. Overrides can be placed on specific devices and trunks
as required. For example, certain wireless networks handle RTP streams using larger (e.g. 40 ms) packets better
than streams using smaller ones.
37