[ih] DNS history and design

[Jeremy C. Reed]

On Tue, 20 Jul 2021, John Gilmore via Internet-history wrote:

> It's likely that a DNS requester that hasn't changed since 1985 could
> plug in and work fine 36 years later, on the Internet of today.

Coincidently, started experimenting with (almost) this a few days 
earlier running 4.3BSD-Reno BIND named 4.8.3 (June 1990) as a caching 
resolver and authoritative service in today's Internet.

It mostly works.

https://dnsinstitute.com/research/2021/ancient-1990-bind-4.8.3.html

Here are two examples of incompatibilities from late 1980s and today:

Using it highlights one standard problem with modern implementations 
that wrongly send EDNS OPT pseudo-records responses back even when 
were never requested.

Also recent implementations are often configured to not respond with 
Additional Section glue and the old named 4.8.3 gives up with SERVFAIL 
at a few levels of recursion, even while its outstanding queries are 
still in progress to find nameserver addresses (and the old named will 
soon answer correctly for same queries).