Why DNS Is So Important to NNM

NNM discovers devices via the IP address found in ARP caches, seed files, routing tables, and ICMP echoes. In IP networks, devices are given names, and in large networks this name space is divided into a hierarchy of subdomains. This eliminates name collisions on large networks where there is a high degree of configuration checking done, generating a high rate of DNS forward and reverse lookups. This lookup mechanism can be a performance bottleneck.

A common problem is NNM combining two routers into a single node. This is usually caused by bad DNS data.

DNS is an efficient mechanism for looking up IP addresses and names because it was designed for that purpose. Caching is one of the mechanisms DNS servers use to create efficiency.

All large companies with large networks depend on their DNS implementations to distribute the authority of names and IP addresses to smaller, more manageable authoritative subdomains. Note that the web-based services on the Internet would not function at all without DNS. Imagine constructing HTML with hard-coded IP addresses. DNS scales very well to the largest networks.

A large part of network management involves routers with multiple interfaces, each with one or more IP addresses. High availability servers usually have multiple IP addresses and interfaces. Devices with multiple interfaces are said to be multihomed. DNS is specifically designed with multihoming devices in mind. In addition, the NNM manual recommends DNS.

You can use NNM’s ovtopodump command to ferret out devices that don’t have DNS entries:

					ovtopodump -Lr > report_name
				

By doing so, you get a very useful one-line summary for each device in the NNM database. On a UNIX system you can pipe the output of the ovtopodump command into one or more filters (such as sort) for some specific purpose. For example, suppose you want to find out which devices in the management domain don’t have DNS entries. Simply do the following:

					ovtopodump -Lr | sort > report_name
				

and examine the entries near the top of the sorted list. The numeric selection names percolate to the top of the list, neatly identifying the devices without DNS entries.