Maintenance Manual

Chapter 2 Fault Tolerance
Network Interface Controllers
2-16
Cisco ICM Enterprise Edition Administrator Guide Release 6.0(0)
• A backup tape taken from the on-line side of the Central Controller database.
At the time of the state transfer, any missing configuration data will be restored.
Historical data is restored by the Recovery process, which is run automatically
each time the Node Manager process starts on the Logger, or by loading the data
from a backup tape.
Software Failure
Cases of software failure that leave a Central Controller database unavailable are
handled in the same way as a disk failure (if the failure cannot be repaired by
existing software tools). Contact your ICM support provider if such a failure
occurs.
Network Interface Controllers
The NIC has four physical controllers on each side of the Central Controller. Each
of these controllers can simultaneously handle calls from the signaling network.
Typically, each physical NIC handles part of the total call routing load for ICM
software.
The NIC processes are implemented as non-synchronized process pairs (one
process on each side of the Central Controller). The NIC runs as a process on the
CallRouter machine.
As a non-synchronized process pair, the NICs operate without knowledge of each
other. They are managed by the Node Manager and communicate with other
CallRouter processes via the Message Delivery Service (MDS). Figure 2-6 shows
how fault tolerance is implemented for various NICs.