Choosing a host naming convention

Whether realizing it or not every organization goes through the process of selecting a naming convention. Often this is done without much forward planing resulting in unwieldy host and service names. In the worst cases host names are dropped completely and either by habit or lack of a proper DNS service only ip addresses are used.

In the following I will talk about the different types of naming conventions that I’ve experienced or read about. I’ll discuss the pros and cons of each. Before we get to that we’ll need to have a small refresher on domain name services.

DNS review

Domain name service is probably the most important and most overlooked of all network services. In many cases unexplained network problems can be attributed to improper or missing DNS. Some view DNS as complicated voodoo that is more trouble than it is worth.

The goal of DNS is very simple. Ip addresses are difficult to remember, this is especially true of ipv6 ip addresses. DNS allows you assign easy to remember names to associate with an ip address. DNS allows further fine grained identification through specialized name and ip address matches.

1. A record: this is the most basic record that associates a DNS name with an ip address.
2. PTR: Sometimes called a pointer or reverse record this associates an IP address with DNS name allowing a person or service to discover the DNS name of a known IP address. When building DNS records I often see these records missing and deemed unimportant. Without these a DNS service will not be complete and errors or service problems may occur.
3. MX RECORD: The MX record identifies the IP address and DNS name of mail servers (mail exchanger). Without these records mail will not flow.
4. NS RECORD: This identifies the IP address and DNS name of DNS servers. Without these records a DNS service cannot function.
5. CNAME: This record, sometimes called an alias, is optional but very useful and often overlooked. A CNAME allows for a host to have additional name records. As we’ll see being able to refer to a host via multiple names can be very useful.

DNS versus host name

It is important to differentiate between a DNS name and a host name. A DNS name is the name record of an IP address. A host name is the name that a host is assigned when it is setup, or changed to. The host name is known only to the host. DNS names and host names independent. They do not always match and there is no guarantee that they will. This is why a proper DNS service is so important. Without one there is no centralized way to refer to a host without knowing its IP address.

Naming conventions

Purpose

This style names host after there purpose.

Examples

Printer1, router1, www and exchange1.

Pros

These host names are easy to remember.

Cons

1. As the number of alike services increases it can be difficult to remember if you are referring to printer1 or printer8.
2. Servers often have multiple purposes. Should a file and print server be called printer1, file1 or print-file1?
3. If a print server is to become a file server you’ll have to go through the trouble of renaming the server.

Geographic

This style attempts to chose names base on location such as city, office or even floor.

Examples

Toronto1, kingst1, and 2ndfloor2.

Pros

You know where to find the server.

Cons

1. What does the host do?
2. As the number of hosts at single location increase it can be difficult to remember if you are referring to toronto1 or toronto5.
3. Whenever a host is moved you’ll have to rename the server.
4. If you’re not using CNAMEs how will your customers refer to the host? Will you give them a new name if you move it?
5. This may give out more information about your host than you might wish.

Inventory

This style attempts to combine the purpose and geographic style in hopes of gaining an inventory system for all hosts. Some go even further to include the operating system in the host name.

Examples

tor-print01, 2ndflr-router02, kingst-exchange01, toronto-lnx-nfs1, kingst-csco-router1

Pros

This hostname tells you a lot about the host.

Cons

1. Host names are long, difficult to remember and difficult to type. This is counter to the purpose of DNS.
2. What if a host has multiple purposes such as file and print?
3. You’ll have to rename a host each time it is re-purposed or moved.
4. While this might seem like a poor man’s inventory system it is no substitute for the real thing.
5. This may give out more information about your host than you might wish.

Themes

We’ve all seen hosts with whimsical names that seem to serve no purpose.

Examples

zeus, maple, blue, pluto

Pros

To the newcomer such host names seem meaningless. However, as they are short they are very easy to remember and type. People will quickly learn to associate such names with their purpose. This is a common technique for code names in industry and the military. These types of host names are the only ones that I can remember even long after I’ve stopped working with them. Further You can never run out of unique names.

Cons

1. The host name tells nothing about the host. One might argue this as a pro since a cracker has no additional information.
2. What happens if a host is decommissioned? Are customers told to access a new host name?

CNAMEs

Using CNAMEs allows you to combine all of the above naming conventions in a single flexible system. Since a CNAME is an alias it can be reassigned to another host without either host requiring a host name change. This allows services to be moved without customers being aware of it. A customer can be pointed to a new web server simply by assigning the CNAME www to another host. The change is transparent.

Considering what you now know about CNAMEs. Go back and look at our naming styles again. We can now combine them in a much more useful way. A host simply called blue can have CNAMEs printer1, tor-printer or even blue-lnx-toronto-printer1. For administrators blue is easy to remember and type. Even if blue is decommissioned we simply reassign the CNAME to its replacement server. If the server is moved, installed with a new operating system or assigned a new purposed simple change the CNAME.

Examples

blue, blue-lnx-toronto-printer1, mars, mars-csc-kingst-brouter1

Pros

1. Allows all of the benefits of the previously discussed conventions.
2. Allows for the movement of services and hosts without the need to change host names.

Cons

1. Requires a functioning DNS service.
2. This may give out more information about your host than you might wish.
3. Current DNS implementations prevent clients from asking a DNS server to list the CNAMEs of a given A record. Instead the DNS server’s configuration must be examined in order to determine all CNAMEs.
4. Requires a little more work to keep CNAME records in addition to A and PTR records.

TXT and HINFO

There are two little known DNS records called TXT and HINFO. These are mostly free form records used to store arbitrary host information. Unlike CNAMEs one should be able to query these fields via a client. However, like some of our other methods this can sometimes offer too much information to a cracker. For more information about the TXT and HINFO refer to RFCs 1464 and 1033.

Whatever method you choose, and this list is by no means complete, be sure to plan ahead. Think about how your naming convention can adapt to a growing network as you add more nodes and move existing ones.