[dns-operations] Any surveys about EDNS0 usage?

[Joe Abley]

On 2011-03-07, at 17:24, Michael Graff wrote:

> My take on this is that it's very hard to measure.  It's very easy to
> tell if someone allows TCP, but EDNS0 support, unless specifically
> queried for and then done carefully, seems hard to measure.

I think you're talking about EDNS0 capability, rather than EDNS0 usage.

Wouldn't it be a reasonable measure of usage to see how many inbound queries to (say) a set of root servers over a reasonable period included the OPT pseudo-RR, and how many did not?

I appreciate there are other things to measure, but that (relatively simple) exercise would give an order-of-magnitude answer to the usage question. Shane and I did something like this a few dnsop meetings back, but it wasn't particularly exhaustive and we didn't check our work.


Joe