Maintenance Manual

2-3
Cisco ICM Enterprise Edition Administrator Guide Release 6.0(0)
Chapter 2 Fault Tolerance
Architecture
primary or backup. Both process sets run in a synchronized fashion, processing
duplicate input and producing duplicate output. Each synchronized system is an
equal peer. Each set of peers is a synchronized process pair.
In the event that one of the synchronized processes fails (for example, a
CallRouter goes off-line), its peer continues to run. There is no loss of data and
calls continue to be routed. When the failed member of the pair returns to
operation, it is resynchronized with its peer and begins to run again as a
synchronized process. Figure 2-1 shows how synchronized execution and hot
standby are applied in ICM software.
Figure 2-1 Duplexed ICM Fault Tolerance
PGs and NICs use the hot standby approach to fault tolerance. Note that the
duplexed NIC in Figure 2-1 is implemented on two separate computers. Each
computer has active and idle connections to the sides of the Central Controller.
NIC fault tolerance is described in more detail later in this chapter.
Duplicated Communication Paths
Each NIC, Peripheral Gateway, and Admin Workstation has two communication
paths to the Central Controller (see Figure 2-1). The two paths connect the device
(for example, a PG) to a Central Controller Agent process on each side of the
37158