APNIC Chief Scientist Geoff Huston Explores the Evolving Puzzle of DNS Repeat Queries and Protocol Efficiency on PING Podcast

In the latest episode of the PING podcast, APNIC Chief Scientist Geoff Huston joins the host to continue a deep and technical discussion regarding the persistent issue of repeat queries observed across APNIC Labs experiments. Over time, the scale and frequency of these redundant requests have shifted significantly, influenced heavily by the continuous evolution of Domain Name System (DNS) protocol usage across the global internet.

Today’s network operators and researchers contend with an entirely different landscape compared to the early days of these observations. Modern DNS architectures feature records that simply did not exist when these anomalous query behaviors were first documented. As a result, systems frequently find themselves generating multiple queries for a wide array of distinct items, creating a complex web of traffic that continues to capture the attention of internet researchers striving to understand underlying infrastructural inefficiencies.

Investigating the Art of Saying ‘No’ in the DNS Ecosystem

A primary focus of Geoff Huston’s recent exploration centers on examining how different methodologies for answering "no" within the DNS framework directly impact query volume. Furthermore, these variations influence the precise amount of processing work required across the entire chain of internet infrastructure, stretching from authoritative servers and intermediate resolvers down to the end clients.

The empirical findings derived from these experiments are frequently surprising to network engineers. Implementing distinct approaches to delivering a negative response—ranging from explicit error messages to withholding a response entirely—yields starkly different operational outcomes. When servers choose not to respond at all, the downstream reaction from clients and resolvers can trigger cascading retry mechanisms, paradoxically increasing the overall load rather than mitigating it. Understanding these nuances is critical for optimizing how the internet handles non-existent domains and missing records, as minor adjustments in server behavior can have amplified repercussions on global network traffic patterns.

Re-evaluating Transport Protocols: UDP, TCP, and State Management

Beyond the semantics of negative responses, Huston’s ongoing research delves into the broader implications that alternative DNS transport protocols might have on query repetition and overall query load. Historically, the architecture of the DNS heavily favored the User Datagram Protocol (UDP) due to its connectionless nature, lightweight design, and minimal overhead. For decades, many network architects and protocol designers argued that the computational and memory overhead required to maintain active connection states for protocols like the Transmission Control Protocol (TCP) made them entirely unsuitable for handling the massive, high-speed scale of the global DNS.

However, as the volume of unnecessary, repeated queries continues its relentless upward trajectory across the modern internet, long-held assumptions are beginning to face fresh scrutiny. Huston’s exploration raises a provocative technical question for the networking community: when evaluated from an end-to-end perspective, could the systematic maintenance of protocol state actually require less computational and network work than repeatedly handling floods of redundant, stateless queries?

While UDP avoids the initial handshake overhead, the sheer multiplication of lost, retried, or redundant packets under adverse network conditions can accumulate a hidden tax. As query repetition scales up, the resources consumed by stateful connection management might eventually prove more efficient, or at least more manageable, than absorbing endless waves of unsolicited or repeated traffic. This ongoing debate touches on core tenets of internet engineering, balancing the desire for stateless simplicity against the practical realities of modern traffic congestion and protocol abuse.

The broader implications of these findings extend far beyond academic curiosity, offering vital insights for organizations and researchers working to maintain the resilience, security, and scalability of the foundational protocols that underpin global digital communications. Readers interested in a deeper dive into the technical complexities of negative DNS responses and protocol behavior can find additional commentary and detailed breakdowns published on the APNIC Blog.

For those wishing to follow the ongoing conversations hosted on the PING podcast, episodes are regularly streamed and made available for subscription across major audio distribution channels. The APNIC community continues to welcome engagement, insights, and research contributions from professionals across the internet operations and networking sectors. Individuals interested in sharing their own findings, proposing topics, or providing feedback on the podcast and blog content are encouraged to reach out directly to the production team via email to help drive future discussions forward.