Maintenance Manual

2-21
Cisco ICM Enterprise Edition Administrator Guide Release 6.0(0)
Chapter 2 Fault Tolerance
Fault Tolerance in Integrated Deployments
Fault Tolerance in Integrated Deployments
Some components in the ICM implement synchronized fault-tolerance, meaning
that communication paths to redundant components are utilized simultaneously.
This reduces the probability of message loss during transition periods to a very
low rate. Cisco Collaboration Server (CS) and Cisco Media Blender (MB) nodes
do not implement synchronized fault-tolerance; however, duplexed
implementations of these nodes provides a vast improvement over a single path
setup. Table 2-1 describes general recovery behavior and its effects when a single
node failure occurs in an integrated deployment. Although multiple failures are
not considered, an instance of multiple node failure will manifest itself as a
superposition of single failure cases. However, a catastrophic failure can occur if
all redundant components in an array fail (for example, if all routing CSs fail).
Table 2-1 Integrated Deployment Failure Recovery
Point of Failure Recovery Action State Lost State Recovered
Effect on Web
Caller
Routing CS
Node
LocalDirector routes to
other CS
Current routing
sessions
None Connection
fails, must
reestablish
Routing MB
Node
CSs activate gateway
connections to other
MBs
None Current routing
sessions
requeued by CS
Time delay
Web PG Node MBs await connection
from other PG web side
which becomes active
None Current routing
sessions
requeued by CS
Time delay
An Agent CS ICM software routes
around this site to
another agent site
Current agent
sessions at this site,
some new sessions
may be lost
Other sites not
affected
Connection
fails, must
reestablish CS
sessions