Summary

  • At RIPE 93, Giovane Moura asked other top-level-domain operators to publish DNS traffic statistics in the same RSSAC-002 format, so researchers could compare them more easily.
  • SIDN’s .nl site publishes four of the six metrics in daily YAML files, with graphs and a methodology page.
  • SIDN’s repository says the unique IPv6 source count has no reconstructable values before 6 April 2025; it also lists 31 March 2026 as a missing day.

The most useful line in .nl’s DNS archive may be the one admitting what is missing. SIDN says its earlier IPv6 unique-source counts were calculated incorrectly. It removed those values because the historical raw data needed to recompute them are no longer available.

That is the practical edge of Giovane Moura’s short RIPE 93 appeal. The Data Scientist at SIDN Labs told the DNS Working Group that TLD traffic statistics had been published in different formats and at different frequencies. Researchers could find figures, but comparing them meant first translating every operator’s file and metric choices. Moura urged other TLD operators to use RSSAC-002, a published measurement specification, as a common format.

SIDN’s public .nl archive gives that proposal a working example. It offers daily YAML files, graphs and methodology for aggregated DNS traffic reaching the authoritative .nl name servers. The site says it covers four of the six RSSAC-002 metrics. Moura said the series reached back about a year and a half and invited other TLD operators to publish their own data.

A shared format removes one kind of friction: researchers can look for the same named measurements in the same file structure. It does not make the underlying histories equally complete. SIDN’s repository marks 31 March 2026 as a day with no data. It separately says num-sources-ipv6-aggregate is available only from 6 April 2025 onward. Before then, the calculation was wrong; the old raw inputs are unavailable, so SIDN cannot produce corrected values. The README says other metrics are unaffected.

That distinction matters to anyone studying IPv6 adoption or changes in the population of visible DNS sources. A chart can share a schema and still begin at a different date, skip a day or carry a correction that cannot be applied backwards. The source file tells readers where the boundary is. That disclosure is part of the dataset, not a blemish to crop out.

RSSAC-002 defines measurement parameters and an interchange format. The hard archival work starts around it: preserve the metric definition, first available date, gaps and correction history beside each series. If the raw inputs cannot be kept, the published record should say so plainly and retain the old calculation’s version history. More TLD datasets would give researchers more material; comparable histories would make that material worth comparing.

Moura’s invitation was simple: publish the numbers in a format other researchers can use. The .nl archive adds a necessary companion rule: publish the limits of those numbers with them.

Sources