Share single DHCP database

The database is designed not to be shared.  You could point it to a third server, but if the 3rd server goes down, your DHCP is down.  So you've only gained complexity and not really any more reliablility.  The best way to do this is to leave the local databases and use the 80/20 rules and have one DHCP server server up 80% of the subnet in each scope and the other servering the other 20% of the addresses.  

See this:
http://technet2.microsoft.com/WindowsServer/en/library/75cd0e1f-f464-40ea-ac88-2060e6769f331033.mspx?mfr=true

If you have clustering you could do this with shared disk.

Here's the 80/20 rule info, it refers to 2000, but still valid for 2003:

http://www.microsoft.com/technet/prodtechnol/windows2000serv/reskit/cnet/cncb_dhc_ogjw.mspx?mfr=true

Our network is too large to use a 80/20 setup.  Plus, my bosses dont want to split it up.  I have about 433 Scopes, 9000 or so clients.  Some scopes are near capacity.  I will have the database backed up somewhere else.  But with the network so reliant on DHCP, i can afford zero downtime.  Cluster will be in place, but i need to know how to share the database (Special accounts if needed, how to point to database, permissions to what).  I agree with you on the complexity, but this is what i need.  Thanks for the help.

You mentioned cluster will be in place?  Are you planning on putting this on a Windows Cluster?  Because if this is what you are trying to do, this is easy.  We do this with our DHCP and it's about the same size as yours.

Do you need a hand configuring that?

Yes, windows clustering will be in place.  I just want to get the database issue resolved first.  I know there are some permissions and reg keys that need changed, but im not sure as to which ones.

Well if you are using Windows Clustering, you put the the DHCP on the shared disk that belongs to the DHCP Resource Group.  Then in the DHCP service parameters in the cluster administrator, you point the database path to your shared disk in the DHCP resource group.

If you are migrating to a cluster from a standalone, during cut over time, turn off the old dhcp service, copy the database to the cluster shared disk.  Then point the DHCP service parameters to the folder that contains the dhcp on the shared disk.  You can also keep the same IP address as your old server (this way you don't have to update every single IP helper on your routers).  Turn on the service and it should pick up the database seamlessly

But, can i do this without cluster inplace.  I can not config clustering yet.  Just want to see if i can get the database working with the 2 machines.  You have been a great help so far.  The points are yours.

Sounds about right.  Thanks for the help.  Im sure i will post more on this issue in the future.

Not a problem

still not working.  If i start the DHCP service with a network account, i can not access the DHCP console.  It says im not auth (but i am).  If i start the service with my network account, i still cant get in to dhcp.  Ideas?

Are you in the local Admin of both servers as well as in the local DHCP Administrators Group?

Yes, im in both.

What are the permissions on the share\NTFS?  Try bumping that up to Full Control.

What happens if you move the database back to the local server?  
I know we're going backwards, but it will narrow things down a bit.  This setup might not work as the DHCP might be internally coded to only use the SYSTEM account.  With a cluster the shared disk is considered local and the SYSTEM can access it fine

Yea, i think ill have to find a SAN.  I believe we have one thats not used.  Ill post a new question (with points) if i need help on the config.  Thanks for all you help.

Fair enough.