In the world of Data Centers equipment is taken down all the time for various reasons. When the PI system is used to monitor a global organization the monitoring of the equipment through the SNMP interface becomes even more problematic.
Case in point:
- Over 90,000 SNMP PI Points
- Over 7,000 unique IP Addresses
- Over 13,000 AF Elements
- Over 600 AF Templates
- Other protocols, such as Modbus, PI to PI, BACNet, OPC, UFL, SNMP Trap and ODBC Link
Managing the PI environment has caused us to spend considerable time on Standards for naming of Elements, PI Points and Templates. This management becomes even more complex when considering the Moves, Adds and Changes that happen often within data centers, which becomes even more complex with a global organization that has over 80 data centers and over 900 other technical spaces world wide.
The current SNMP interface will shutdown a Scan Class and the whole interface when the volume of IP Addresses do not respond (I/O Timeout) reaches a certain percentage. This will have a dramatic affect on the operational IP Addresses in the same Scan Class and Interface.
What we need is an SNMP interface that dynamically allocates a unique Scan Class per IP Address and then be able to tell the Interface to ignore IP Addresses that go into I/O Timeout while not reducing the scan frequency of the other operational IP Addresses.