Theory question about SRV records and Authentication

[Cstorms]

Is this possible...

Setup 2 domain controllers, not only for failover but using both for authentication and do this by using a lower priority designation on the GC SRV record (i think this is the one that you would set..??) and put it at like 50 on both machines, setup a ttl on it to something like 0 or whatever and then when a machine tries to authenticate there would be a 50/50 chance of it using either of the servers...

I have no idea if I am way off base on my thinking here, but would this not be a good idea for load balancing for heavy load times in the morning? I know that I could setup a seperate site and maybe just vlan a portion of the network off and do it this way but I wanted to verify the craziness of this idea first.

Any other methods out there in use for recommendation are appreciated, also, what types of load are you seeing on your production enviroments when users first login in the morning (the big rush) and how many users are in your enviroment.


Thanks!

 

[Cstorms]

Alright in response to my own question in case anyone else wanted to know..

If you were to lower the priority weight (1-100) on the 2 service records that are queried via the netlogon process, it would be essentially like using percent as the corresponding terminology... ie... _LDAP._TCP.dc._msdcs.domainname with a priority weight of one record of 50, and that of the other server being at 50, it would be like flipping a coin with no overall consistancy.

Once queried it would then stay accurate until a manual clearing of the dns cache or a ttl is reached and the record expires in which case a UDP broadcast is sent and the response is repeated.

So to answer my own question, one would have to say that aside from setting up seperate sites there would be no exact science for determining which dc a client with authenticate to.

Cheers