Hi Hugo, On 21/02/2017 15:16, Hugo Salgado-Hernández wrote:
Dear Chris. It seems fine from my perspective. The second measurement should be answered by resolver's cache, being top-resolved names.
But what are you thinking on the first measurement? I'd try to reduce to a minimun sending random queries to root servers, an infrastructure frequently abused :( Also for the benefit of transparency and politeness you could prepend some string on the form "ripe-atlas-test-measurement.<random>."
The intention of the first measurement is to check the actual reachability of the root zone name servers from probe resolvers. This is a measurement which was requested by the K-Root operations team within the RIPE NCC, and is intended to show the reachability of the root zone servers in general given issues with a particular root letter. These measurements will indeed be constructed in a way similar to how you suggested: <probe_id>.atlas.ripe.net.<random>. Kind regards, Chris Amin RIPE NCC