Controversial SOA values ?

[Jim Reid]

>>>>> "Michael" == Michael Milligan <milli at acmebw.com> writes:

    Michael> If you can't get your master back online within a week,
    Michael> something is horribly wrong with your service planning...

I agree and disagree with this. It should be possible to get a master
server back on-line within a week.

OTOH, circumstances beyond your control could make that an unrealistic
target. Maybe the telco has to run in a new cable. [I once saw a
trouble ticket which said the time to repair depended on the
weather. A ship had to haul up the faulty cable from the sea bed in
the middle of the Atlantic.] Perhaps there's a fire or flood - or
even an earthquake! - in the computer room. Or the relevant PHBs are
not available to authorise change requests or requisition new
hardware. Big companies can be especially prone to bureaucracy like
this. Maybe the failure takes place when the whole country is on
holiday? Lots of offices in Europe shut down for 8-10 days over Xmas
and the New Year.

It all depends on your assumptions about the nature of the disaster or
catastrophic failure and the plans for recovering from that. One place
I used to work developed a recovery plan for a jumbo jet parking
itself in the computer room, destroying all the hardware and backups
as well as killing everyone in the building. [The office was 2-3km
from the end of a runway at a major airport.] If something as extreme
as that happens, getting the master name server back in a week might
not be feasible. Of course availability of the DNS might not even
matter by that point.