Why forwarding is a Bad Thing

[Jim Reid]

>>>>> "Brad" == Brad Knowles <brad.knowles at skynet.be> writes:

    >> The larger cache of the forwarded server can sometimes make
    >> lookups faster, but not enough to make a significant
    >> difference.

    Brad> 	I disagree.  If the query is one that would take a
    Brad> while to resolve, you could cause a mail message to be
    Brad> deferred (because the lookup doesn't resolve in time)
    Brad> instead of potentially being delivered, and depending on
    Brad> what else happens on that server in what order and at what
    Brad> time, a single deferral could potentially result in the mail
    Brad> message not being retried again for several hours (or more).

We can agree to differ. I recognise the problems that latency can
cause on a high-throughput system, especially a mail server. You have
much more experience and insight into that than I do. However I don't
accept forwarding name servers make a significant difference in this
scenario. You're always going to get deferred lookups because mail
will get delivered to domains that have badly broken or dead name
servers anyway. I would have thought that the way to deal with that is
engineer the mail system to chuck mail for delivery to tardy servers
on to a system set up for that? ie If it can't get an answer from the
DNS and open a connection to port 25 in N seconds, throw the mail at
an overflow system that is set up to cope with a deep queue of mail
bound for "slow" servers. "If I can't deliver this mail *right now*,
get rid of it to somebody else so my fast queue doesn't back up."