I don't see the point in making that restriction. You'll get all the SRV
records in one query anyway; all you have to do is deal with them if
more than one exists. That really isn't hard, and it's the expected
normal behaviour of DNS records. I see no reason to deviate from that.
But you have to loop through them to handle the A record that each can
point to. This changes the number of lookups from 2 to N+1, where N
is the number of SRV records. Since SRV records are pretty small, you
could fit 50 of them into a UDP DNS packet without any trouble, and I
think that a scheme that can require 50 lookups where people are
expecting 2 would be bad.
I'm not saying that the extra complexity of multiple SRV will break
this camel's back, but you'll notice that it already encouraged
someone to propose adding yet more cruft to describe IP ranges, to
which the answer is good lord, no. That's both for complexity reasons
and because the intention of CSV is to have a CSV record per client
machine in preference to an omnibus mess that covers hundreds or
thousands of them. I really think that it is to our advantage to
keep CSV as simple as possible, and I just don't see that multiple
SRV offers important utility that's worth the performance hit of
going from O(1) to O(N).
R's,
John