The MGCP gateway maintains a remote connection to a centralized Cisco Unified Communications Manager cluster by sending MGCP keepalive messages to the Cisco Unified Communications Manager server at 15-second intervals. If the active Cisco Unified Communications Manager server fails to acknowledge receipt of the keepalive message within 30 seconds, the gateway attempts to switch over to the next available Cisco Unified Communications Manager server.
If none of the Cisco Unified Communications Manager servers respond, the gateway switches into fallback mode and reverts to the default H.323 session application for basic call control. H.323 is a standardized communication protocol that enables dissimilar devices to communicate with each other through use of a common set of codecs, call setup and negotiating procedures, and basic data transport methods. The gateway processes calls on its own using H.323 until one of the Cisco Unified Communications Manager connections is restored.
The figure below illustrates a typical VoIP network topology in which MGCP gateway fallback is supported.
Figure 2. Typical VoIP Network Topology Supporting the MGCP Gateway Fallback Feature
The MGCP Gateway Fallback feature provides the following functionality:
MGCP gateway fallback support--All active MGCP analog and T1 CAS calls are maintained during the fallback transition. Callers are unaware of the fallback transition, and the active MGCP calls are cleared only when the communicating callers hang up. Active MGCP PRI backhaul calls are released during fallback.
Any transient MGCP calls (that is, calls that are not in the connected state) are cleared at the onset of the fallback transition and must be attempted again later.
Basic connection services in fallback mode--Provides basic connection services for IP telephony traffic that passes through the gateway. When the local MGCP gateway transitions into fallback mode, the default H.323 session application assumes responsibility for handling new calls. Only basic two-party voice calls are supported during the fallback period.
Except for ISDN T1 and E1 PRI calls, all the MGCP calls that are active at the time of fallback are preserved, but transient calls are released. When a user completes (hangs up) an active MGCP call, the MGCP application handles the on-hook event and clears all call resources.
When the fallback mode is in effect, the affected MGCP gateway repeatedly tries to open a TCP connection to a Cisco Unified Communications Manager server in the prioritized list of call agents. This process continues until one of the Cisco Unified Communications Manager servers in the prioritized list responds.
The TCP open request from the MGCP gateway is honored, and the gateway reverts to MGCP mode. The gateway sends a Restart-in-Progress (RSIP) message to begin registration with the responding Cisco Unified Communications Manager.
All currently active calls that are initiated and set up during the fallback period are maintained by the default H.323 session application, except ISDN T1 and E1 PRI calls. Transient calls are released. After rehome occurs, the new Cisco Unified Communications Manager assumes responsibility for controlling new IP telephony activity.
The following types of interfaces on the gateway are supported:
FXS analog interfaces--For connecting to the PSTN or analog phones
FXO analog interfaces--For connecting to the PSTN or PBXs
T1 CAS digital interfaces--For connecting to the PSTN or PBXs
T1 and E1 PRI digital interfaces--For connecting to PBXs and central offices (COs)