Polling architecture

Use this information to understand how the components of the Network Manager polling process work together to poll network devices.

The Polling engine, ncp_poller, is the component that controls network polling. The Polling engine uses active polling operations to gather data. This data is used to trigger alerts if appropriate, and can optionally be stored in the NCPOLLDATA database for later analysis in the performance reports.

Note: By default, there is are two instances of ncp_poller running on the Network Manager server:

ncp_poller_default
ncp_poller_admin

Therefore by default there are two instances of ncp_poller running on the Network Manager server, each with a different service name, as specified above.

After you create polls and poll definitions, they are saved to the NCMONITOR schema within the NCIM database. The ncp_poller process reads the poll definitions from the NCMONITOR schema. The following figure shows how ncp_poller interacts with other Network Manager components and the ObjectServer. Note that the figure does not include the historical data storage feature.

Polling dataflow — Figure 1. Polling data flow

1 Polling targets are retrieved: Network Manager has a default set of polling policies. These polling policies include simple device or interface pings and more complex threshold polls against specific MIB variables and are stored in the NCIM database. The Polling engine, ncp_poller, retrieves polling policies from the NCIM database and determines the target devices to poll, and how to poll them. Network administrators can log into the Polling GUI to configure these polling policies. For example, they can configure the system to poll a more restricted set of devices, to change polling frequency, to change the data collected.
2 Network devices are polled: The Polling engine, ncp_poller, polls network devices based on the polling policies defined in Network Manager. Devices are polled based on polling frequency and based on the set of devices specified in the polling policy.
3 Relevant poll results are converted to Tivoli Netcool/OMNIbus events: Poll results that are converted into Tivoli Netcool/OMNIbus events include only those polls where the response indicates a device or other network failure of some sort, such as a threshold violation or an ICMP ping fail. The Polling engine, ncp_poller, sends these poll results to the Probe for Tivoli Netcool/OMNIbus. This probe maps the poll results into Tivoli Netcool/OMNIbus event format.
4 Optionally polled data is stored in the NCPOLLDATA database: Depending on poll policy configuration settings, ncp_poller sends collected data to the NCPOLLDATA database, where it is stored and can be retrieved using reports. The NCPOLLDATA database should reside on the same server as the NCIM database for optimal performance.
5 Events sent to ObjectServer: Probe for Tivoli Netcool/OMNIbus sends the converted event data to the Tivoli Netcool/OMNIbus ObjectServer. Tivoli Netcool/OMNIbus and other probes also sent events to the ObjectServer.