Designing and Managing Behavior Models - Using Other Data Sources -
Enabling a Trap Mask      NerveCenter's Built-In Triggers

Using Other Data Sources

For the most part, NerveCenter behavior models detect network and system conditions by using polls and trap masks to poll SNMP agents and respond to SNMP traps, respectively. Thus, a behavior model's main source of information is devices running SNMP agents. However, NerveCenter behavior models can obtain data from other sources as well.

For example, a behavior model on one NerveCenter server can receive information from a second NerveCenter server. The second server uses an Inform alarm action to notify the the behavior model on the first server of a condition it has detected. This Inform action involves sending what appears to be an SNMP trap to the first server. Actually, the message is not an SNMP trap -- it is sent via TCP rather than UDP -- but the behavior model receiving it treats it exactly as if it were a trap.

NerveCenter behavior models can also receive input from Hewlett Packard's IT/Operations. IT/Operations manages a variety of elements: applications, databases, systems, and networks. IT/Operations can use NerveCenter to correlate the conditions it detects. To communicate with NerveCenter, IT/Operations sends messages containing information about detected conditions. On the NerveCenter side, a behavior model reads these messages using a mask that is similar to a trap mask, but is tailored to handle IT/Operations messages. After correlating events detected by IT/Operations, NerveCenter can send a message to IT/Operations using an alarm action called Inform OpC (IT/Operations was formerly called OperationsCenter).

Finally, NerveCenter behavior models can obtain information about network conditions from NerveCenter itself. In particular, when NerveCenter sends an SNMP or ICMP message to a device and the message results in an error (perhaps because the node is unreachable), NerveCenter can notify a behavior model of this condition. NerveCenter does this by using what are called built-in triggers, such as NODE_UNREACHABLE, which can cause state transitions in an alarm just as other triggers do. These triggers are necessary because devices that are down or unreachable cannot respond normally to NerveCenter polls, or send SNMP traps to NerveCenter.

For further information about these additional sources of input, see the following sections:

Section	Description
NerveCenter's Built-In Triggers	Discusses what trigger NerveCenter can fire automatically and how to use these triggers in behavior models.
Another NerveCenter	Explains how a behavior model on one NerveCenter server can inform another server of a condition it has detected.
HP OpenView IT/Operations	Explains how to use an OpC mask to interpret a message sent from IT/Operations to NerveCenter and how to notify IT/Operations of a condition detected by NerveCenter.