Authoritative Domain Name System (DNS) response selection defines query-time response selection based on resolver-visible context and per-answer metadata, yielding different observable outcomes for the same query under different conditions. Although such behavior is widely deployed and often described informally as traffic steering, its semantics have not been formalized independently of particular configuration languages or implementations. This paper shows that authoritative DNS response selection inhabits a bounded semantic domain determined directly by DNS protocol constraints. Requirements such as finiteness of responses, RRset atomicity, termination, cacheability, and restriction to resolver-visible inputs jointly limit the expressive power of any query-time selection mechanism. We formalize authoritative response selection as a class of DNS-admissible functions and prove that every such function admits a finite normal form consisting of conditional restriction over observable context followed by selection among a finite candidate set. We further show that this bounded semantic domain carries intrinsic algebraic structure induced by DNS semantics, enabling principled reasoning about composition, expressiveness loss, and semantic collapse. Concrete authoritative systems, configuration models, and serialized encodings are modeled uniformly as semantic restrictions of this domain. This framework supports precise reasoning about equivalence, representability, and approximation across heterogeneous authoritative DNS systems, grounded directly in protocol semantics rather than implementation detail.

翻译：权威域名系统（DNS）响应选择基于解析器可见的上下文和每个答案的元数据，在查询时定义响应选择机制，使得同一查询在不同条件下产生不同的可观测结果。尽管此类行为已被广泛部署，且常被非正式地描述为流量引导，但其语义尚未独立于特定配置语言或实现形式化。本文证明权威DNS响应选择存在于由DNS协议约束直接决定的有界语义域中。响应有限性、资源记录集原子性、终止性、可缓存性以及限于解析器可见输入等要求共同限制了任何查询时选择机制的表达能力。我们将权威响应选择形式化为一类DNS可容许函数，并证明每个此类函数都允许一种有限正规形式，该形式由对可观测上下文的条件限制和有限候选集内的选择组成。我们进一步证明，这一有界语义域承载着由DNS语义诱导的内在代数结构，支持对组合性、表达能力损失和语义坍缩的原则性推理。具体的权威系统、配置模型和序列化编码被统一建模为该语义域的限制形式。该框架支持基于协议语义（而非实现细节）对异构权威DNS系统间的等价性、可表示性和近似性进行精确推理。