Background

As networks have grown in complexity, so have these name resolution mechanisms increased in sophistication.

NetBIOS and DNS Computer Names

Windows NT networking components rely on a naming convention known as NetBIOS. In general, NetBIOS computer names consist of a single part.

In contrast, TCP/IP components rely on a naming convention know as the Domain Name System (DNS). DNS computer names consist of two parts: a host name and a domain name, which combined form the fully qualified domain name (FQDN).

Fortunately, NetBIOS computer names are compatible with DNS host names, making interoperation possible between the two. Windows NT combines the NetBIOS computer name with the DNS domain name to form the FQDN.

Note

Under Windows NT, the DNS host name defaults to the same name as the NetBIOS computer name. You can change this if you need separate names.

Flat vs. Hierarchical Name Spaces

The original naming scheme for both NetBIOS and TCP/IP consisted of a flat name space where each computer was assigned a single-part name. (A single-part name consists of a short sequence of characters without any additional structure.) Flat name spaces worked well for simple networks with relatively few interconnected computers, but as network complexity increased, they rapidly become inadequate for the following reasons:

• Single-part names are derived from a finite set of identifiers. The potential for conflict increases with the number of computers interconnected.

• Administration of the name space rests with a central authority. Someone must ensure that each computer is assigned a unique name. Because there is no convenient way to segment a flat name space, control of name assignment must be centralized.

• All changes to the network must be approved by the central authority. Before a new computer is added to the network or an existing computer is moved to a different subnet, the change must be coordinated with the central authority. For a large network, this results in a significant administrative burden.

A hierarchical name space implemented as a multi-part naming scheme enables authority to be distributed and administration to be decentralized. A hierarchical name space can be viewed as an inverted tree with the branches and leaves pointing down. A central authority still manages the top of the tree, but below the top level the structure can be distributed into autonomous administrative units. Name uniqueness must still be enforced at the lowest administrative level, but this is a reasonable task for a well-segmented name space. The hierarchical structure of the name space guarantees name uniqueness above these lower levels.

Implementations of hierarchical naming schemes exist for both TCP/IP and NetBIOS: Domain Name System (DNS) for TCP/IP and NetBIOS Scope for NetBIOS.

Name Space Implementations

The first implementations of name spaces — both flat and hierarchical — relied on text files for mapping of computer name to IP address. Each computer on the internetwork had its name and IP address on a line in the file, and a copy of the file existed on each computer. This solution worked well for simple networks having relatively few interconnected computers. As networks grew in size and complexity, this method ran into scaling problems similar to those experienced with a flat name space.

Newer implementations have largely done away with the need for a mapping file on each machine; instead, server-based repositories store the necessary information. Mapping files still exist but are typically used in simple networks or as a safety feature in case the name servers are down.

The mapping files are

• HOSTS for DNS names

• LMHOSTS for NetBIOS names

Note

When you install Windows NT, example HOSTS and LMHOSTS files are placed in the \systemroot\System32\Drivers\Etc directory.