Internet hostnames: extended description and recommendations

Overview
The GeneralReport "Internet hostnames: extended description and recommendations" by Daniel J. Bernstein presents a systematic examination of hostname practice and policy as they existed in the early 1990s, then offers pragmatic guidance to reduce ambiguity and interoperability problems. The document treats hostnames not just as labels attached to machines, but as elements of a distributed namespace with both technical and administrative consequences. It emphasizes clarity of syntax, predictable comparison rules, and separation of naming from other uses such as human-readable descriptions or policy signals.

Problems with existing hostname practice
Bernstein documents a variety of practical problems that arise when different implementations and administrators adopt inconsistent conventions. Ambiguities about permitted characters, case sensitivity, label and overall length, and the mixing of local and global namespaces cause failures that are intermittent and hard to diagnose. The interplay among /etc/hosts-style local files, DNS, and other naming services produces divergent behavior on different systems; applications sometimes make unsafe assumptions about numeric labels and dotted-quad addresses, leading to misinterpretation. The report highlights how these small ambiguities accumulate into widespread operational fragility.

Syntax and canonicalization
A central recommendation is to adopt a clear, minimal syntax for hostnames and to mandate canonicalization rules for comparison and storage. The guidance narrows the permitted character set to ASCII letters, digits, and the hyphen, enforces label-length and total-length limits consistent with DNS constraints, and prescribes case-insensitive treatment. Recommendations also address prohibited patterns such as leading or trailing hyphens and discourage labels that are purely numeric to avoid confusion with IP address literals. By insisting on deterministic normalization and rejection of ill-formed names, the proposal aims to simplify parsing, reduce edge cases, and make programmatic checks reliable.

Operational and administrative recommendations
Bernstein urges explicit separation between globally resolvable names and locally meaningful identifiers, recommending that fully qualified domain names be used wherever global resolution is required and that local aliases be kept out of global registries. Delegation of administrative authority and consistent policy for name allocation receive attention as ways to limit name collisions and misconfigurations. The report favors use of DNS as the authoritative distributed mechanism for hostname resolution while recognizing transitional needs for legacy hosts files and advising conservative migration strategies that avoid sudden breakage.

Design principles and longer-term implications
Beyond specific rules, the report advocates design principles intended to prevent future problems: keep the namespace simple, avoid embedding semantic content in hostnames, and minimize assumptions made by applications about naming structure. Bernstein recognizes trade-offs between rigid rules and operational flexibility but leans toward rigidity where ambiguity causes persistent interoperability failures. The recommendations anticipate later debates about internationalized naming and the limits of ASCII-based conventions, signaling the need for future extensions to address non-ASCII requirements without undermining core stability.

Impact and relevance
The recommendations crystallize practices that reduce configuration errors and make name handling predictable across platforms and software. Many of the syntactic constraints and canonicalization behaviors recommended align with or prefigure later standardization efforts and common implementation choices. The emphasis on clear semantics, stable delegation, and conservative transition strategies remains relevant for administrators and protocol designers dealing with evolving naming requirements, internationalization pressures, and the persistent need for robust, interoperable host identification.