>> I understand about callforwards, but it seems to me that the backup MX
>> is mostly going to get mail when the main MX isn't available, so I
>> don't see how this would work very well in practice.
>
> Actually, it works pretty well in practice. The main problem with secondary
> MXes is that spammers send messages to them even when the primary is up (which
> in most cases is 99+% of the time).
This leads to the fairly obvious question: if the primary is up 99+% of
the time, what's the point of a secondary?
The point is, as the saying goes, that "past performance is not an indicator
of future results".
I have several machines here that have operated at three or even four 9s for
years. Doesn't mean I don't back them up regularly, because sooner or later
there's bound to be a catastrophic failure of one sort or another.
There was an era when they
were useful to deal with networks that didn't have routes to some parts of
the net, but I'd think that's a pretty small niche now.
Yes, but asymmetric routing faults are still pretty common, especially in
developing countries. A fairly common setup I've seen is to have a secondary
somewhere else that has a backup link to get mail to the primary when regular
connectively isn't working.
Unless your
primary is down for a long time, days or longer, any legitimate mail will
just wait and retry and get delivered anyway.
Prolonged primary outages are exactly the case having a secondary is mostly
concerned with these days.
And again note that the outages may be network, not server, related. I have no
idea if my experience is typical, but I've seen far more cases where a
secondary became important because of a prolonged network outage than because
of a server outage.
Ned