Subsequent topology discovery and maintenance

After the initial discovery of the network topology, you must choose a schedule for subsequent discoveries to maintain an accurate topology:

• Full discovery should be scheduled to occur at least once per week.

  Full discovery is the discovery of the devices in the IP Manager’s repository.

  Scheduling full discovery during a long period of low network utilization, such as a weekend, is recommended. For example, a large multinational bank discovers devices for a domain on Saturdays at 1 p.m. Specific times are scheduled using cron or sm_sched to invoke sm_tpmgr --discover-all for the domain.

  The Global Console allows you to specify a full discovery interval, which is the time between the initiation of full discoveries. The full discovery interval is counted from the time when the IP Manager is started. If a discovery is in progress when the next scheduled discovery is to begin, the Topology Manager for the IP Manager skips that full discovery and writes an exception to the IP Manager’s log file.

  Note:

  If you use cron or sm_sched to schedule full discoveries, clear the Enable Full Discovery option at the Domain Manager Administration Console.

• Pending discovery should be scheduled to occur at least once per day.

  Pending discovery is the discovery of the devices on the IP Manager’s Pending Devices list.

  Like full discovery, schedule the pending discovery during a period of low network utilization. The duration of a pending discovery is usually much shorter than full discovery, so schedule it during a relatively idle work shift. For example, a regional service provider discovers pending devices on weekdays at 2 a.m. As with full discovery, specific times for pending discovery are scheduled using cron or sm_sched to invoke sm_tpmgr --discover-pending for the domain.

  And, as with full discovery, the Global Console allows you to specify a pending discovery interval, which is the time between the initiation of pending discoveries. The pending discovery interval is counted from the time when the IP Manager is started.

  Note:

  If you use cron or sm_sched to schedule pending discoveries, use 99 days for Discover Pending Interval at the Domain Manager Administration Console.

  Running discoveries more often will provide a more accurate network topology, but you must consider both your needs and the cost in terms of resources. If your network changes more often than the recommended discovery schedules, you must shorten the time between discoveries. Typical network changes include anything from hardware changes such as adding cards or devices, to configuration changes such as reassigning IP addresses or modifying a VLAN.