/ R&D News: Paper on DNS Magnitude published
/ nic.at News - 23.11.2020 07:45
R&D News: Paper on DNS Magnitude published
What sounds at first like a seismology term is actually to do with the Domain Name System: The "DNS Magnitude" has been developed by our Head of R&D Alex Mayrhofer together with his colleagues Michael Braunöder and Aaron Kaplan, and represents a DNS-based metric of the popularity of a domain. The idea for this was developed in 2016 and has now been finalized in the form of a comprehensive paper supported by ICANN (Internet Corporation for Assigned Names and Numbers).
The use of the term "magnitude" is of course no coincidence: Just like the description of the strength of an earthquake, it is a logarithmic measurement, since the popularity of different domains varies dramatically. The use of a logarithmic scale makes it possible to represent this range simply and easily understandablly for humans on a scale between 0 and 10. "Quite simply formulated: The DNS Magnitude allows an estimate of how popular a domain is. The basis for this is the assumption that popular domains are requested more often than less popular ones" explains Alex Mayrhofer.
It is not the number of queries per domain, but the number of unique IP addresses that is decisive for the calculation. According to Alex Mayrhofer, this is necessary because the TTL (time-to-live), i.e. the period of validity of the resolution, varies greatly in the DNS and for this reason a count of the queries is not very meaningful: "The TTL can deliver in any timescale from a few seconds up to several weeks. After expiration, the client must send another query. This naturally leads to domains with a very high TTL having fewer queries. We wanted to exclude this property of the DNS in order to make the DNS magnitude as independent of the TTL as possible".
Since the DNS magnitude considers the entire DNS traffic of a domain (and not only the queries for accessing the website), infrastructure domains are well regarded. "The importance of such domains is often underestimated. But if such a domain, which represents the name servers for tens of thousands of other domains, fails, it naturally pulls all these other domains with it into the abyss - infrastructure domains are therefore often very much underestimated, and DNS Magnitude represents the importance of these crucial domains better than counting accesses to the corresponding web server," explains Alex.
This is why the ranking of .at domains by DNS Magnitude is led by the infrastructure domains of large registrars and providers. If, on the other hand, one filters for queries caused by visits to web servers, google.at leads the field among the .at-domains – as expected - with a DNS magnitude of 9.0; amazon.at is in 13th place with a value of 7.8.
This information can be used, for example, to make recommendations to customers for additional products such as our anycast service RcodeZero DNS, or other security features. Such services make particular sense for domains with a high magnitude, as these are especially critical for the smooth functioning of the Internet.
In the published research paper, Mayrhofer and his colleagues particularly address the use of DNS magnitude in the root zone: "The measure can of course be calculated for entire top-level domains. It is shown that the magnitude correlates well with the size of a TLD - .at, for example, is usually ranked between 33 and 40. It was particularly exciting for us to work with traffic data from the root servers, which is not an everyday occurrence," says Alex.
Even after the publication of the work, Alex Mayrhofer is far from finished with this topic. "We are currently testing internallyhow meaningful the data for individual .at-domains is. For example, whether there is a connection to the likelihood that a domain will be deleted in the near future. And, of course, we also want to make the data available to our registrars, if it is of use to them - and to our domain holders".
You can download the full paper here: