Watching performance on a DHCP Server

[Barr Hibbs]

> -----Original Message-----
> From: dhcp-users-bounce at isc.org
> [mailto:dhcp-users-bounce at isc.org]On
> Behalf Of sthaug at nethelp.no
> Sent: Tuesday, February 12, 2008 13:21
>
> I'm planning for the extreme cases by having a
> redundant server (RAID, redundant power etc),
> and soon DHCP failover. I'm *not* planning for
> our DHCP infrastructure to cope with 100K
> requests at once, since I find such a scenario
> highly unlikely (even considering the failures
> that you have mentioned above).
>

I won't claim to be the sole authority on the topic, but in
the case I reported earlier the enterprise's revenue stream
was directly affected by any outages.  That's why we used
redundant data centers on protected power with a third for
unimaginable emergencies, doubly redundant networking
connections with diversely routed circuits, and used every
best practice for infrastructure survival in earthquake
country...  but when commercial power was lost and restoral
took nearly 3-1/2 days, we really stretched the capacity of
our network and servers.  While simultaneous power restoral
after a wide-spread outage is unlikely, it is absolutely
possible to have a huge in-rush of INIT-REBOOT requests.

--Barr