Fail Over DHCP/DNS

Associate
Joined
19 May 2005
Posts
862
Any 1 point me in the right direction for a guide on how to do fail over DHCP/DNS windows 2003/2008 servers

I know its not easy but i kinda need to do it as its for a LAN with 200 people on it and i fear the day when the dhcp server goes down and every one cant work :mad:

Any help is appreciated

Phil
 
You could just set up 2 dhcp servers with the same config but slightly different ranges.

DNS fails over anyway, if the primary does not respond the client will try to contact the secondary, and so on.
 
Any 1 point me in the right direction for a guide on how to do fail over DHCP/DNS windows 2003/2008 servers

I know its not easy but i kinda need to do it as its for a LAN with 200 people on it and i fear the day when the dhcp server goes down and every one cant work :mad:

Any help is appreciated

Phil

This a problem I've been looking at of late; the DHCP Specifically; we've several DNS Servers.

There are 3 approaches as far as I can tell;

1) 3rd Party Software - Cisco (did?) make a dhcp server software, which could be clustered; and im sure if you looked something similar could be found. Essentially when either server gives out an address it checks with/updates the other; so they both have an accurate copy of the lease database etc.

2) Split Scope - Take your address range and split it down; this is fine if you've large subnets, but more difficult if you've tightly trimmed them. There are some guidelines from microsoft around for doing this; something to do with the number of leases that are likely to expire before you can get a new server running.

3) VM - The way I decided to do it; was to put the DHCP server onto our Xen cluster. This gives us a box that can float between hardware failiures, and is snapshotted every night in case of corruption/os faliure; I can essentially re-provision it in 5-10mins should that happen. Whether this is truely resiliant enough I dont know; but I couldnt really see any recommendations for or against the approach; so i'll see how it goes.
 
Use 2 DHCP servers. If your scope renewal times are long enough a single server outage should not be a problem. Easily enough for a DHCP server rebuild recovery.
 
3) VM - The way I decided to do it; was to put the DHCP server onto our Xen cluster. This gives us a box that can float between hardware failiures, and is snapshotted every night in case of corruption/os faliure; I can essentially re-provision it in 5-10mins should that happen. Whether this is truely resiliant enough I dont know; but I couldnt really see any recommendations for or against the approach; so i'll see how it goes.

Yes i like this method for making services HA. Although if you don't have it already in place it's going to be the most expensive DHCP server EVER :p
 
I'd jsut install 2 dhcp servers with a split scope, I'd hope you have 2 DC's I'd jsut put dns on 2 and ad integrate them

If its only 200 users keep it simple.... we spend more time trying to fix clusters than we do the software on them...
 
Place i'm at now has a clustered DHCP solution for 14k or so desktop PC's over the environment, lots of sites throughout the EU.

Thats right, people in Madrid come to the UK for DHCP.

Crazy idea imo, but MS signed off on the design.
 
Back
Top Bottom