Remember the old management adage "You can only manage what you can measure"? Well, there's a corollary in Network Management, in which you can only manage devices that you know to exist. In my current job, the network has over 600 core network devices (e.g. routers and switches) and in the entire network, there are over 4000 network devices. How do we know that the network management platform has access to all the key devices? Well, we needed a way to check.
I built a report in NetMRI today that would tell us which network devices have been discovered, but are not being managed. The process requires a good network discovery as its foundation. I've used a variety of network management tools and NetMRI's discovery is one of the best I've seen. It relies on multiple sources of data, such as ARP, switch forwarding tables, CDP, and routing neighbors, to discover neighboring devices, then tries to discover the SNMP and login credentials to use for each device. The security team should be notified prior to running NetMRI on the network because any SNMP and login attempt failures should create a security alarm. (If it doesn't, then the security team should be concerned.) Once SNMP access is obtained, neighbor information from the newly discovered device is used to expand the discovery.
How do you find out about devices where SNMP access was not successful? Hopefully, you have a standard way to identify your networking devices, such as having known device management address ranges or known device naming standards. You may need to use a combination of these approaches to identify the network devices. Fortunately for me, the naming standards where I'm working made it easy to identify network devices. So I built a report that identifies devices whose DNS name matches the known naming pattern. In this network, the naming pattern used a site ID, so the report filter was:
1: Devices: DNS Name matches /^[0-9][0-9][0-9]-/
The matching pattern defines a regular expression (the text between the '/'s) that matches beginning of the name ('^') and then three digits ([0-9][0-9][0-9]), followed by a hyphen. This will match any device name that starts with NNN- (where N is a digit).
I then needed a way to test whether NetMRI had SNMP access to each device. NetMRI won't increase the device assurance above 75 without having SNMP access, so I added a second filter rule:
2: Devices: Assurance < 90
The two filter rules were combined by a logic definition that required both conditions to be matched:
Logic: 1 and 2
This report did the job and quickly identified about ten devices that had been discovered but which NetMRI could not access. These devices didn't have the correct SNMP configuration, and having the list made it easy to quickly correct them.
-Terry