DHCP Server & DHCP Relay Agent
On Win2k Server can we configure DHCP Server & as well as DHCP relay agent on the same Server.
if not why ?? could you explain how a DHCP Relay Agent works ?
Windows Forum
On Win2k Server can we configure DHCP Server & as well as DHCP relay agent on the same Server.
if not why ?? could you explain how a DHCP Relay Agent works ?
by duskos
DHCP Relay Agent is used so that clients can 'connect' to DHCP server which is located in another subnet (accross a router). Normaly, clients 'connect' to DHCP server by using special broadcast packet (DHCPDISCOVER). However, if DHCP server is on the other side of the router, this brodcast packet will not pass thru the router (unless the router is configured to do so, i.e. is RFC1542 compliant). This is were DHCP Relay Agent comes into play. It listens for DHCP type broadcasts (DHCPDISCOVER, DHCPSELECT etc) on the local subnet and forwards them to the DHCP server accross the router. The DHCP server accross the router sends replies to DHCP Relay Agent. DHCP Relay Agent then broadcasts (relays) these replies to the client on the local subnet.
So, the answer to your question is:
There is no point to install DHCP Relay Agent in the same subnet as DHCP Server (or on the same server) because in that case client can reach DHCP server directly, without a help from DHCP Relay Agent.
by k4k
by jaystar
While I agree with the above statement on HOW the process works, I would like to raise a point of why you may want both a DHCP Relay Agent AND a DHCP server on one segment. If you follow the generally approved rule of leasing 80% of your available IPs from one server, and have a second server that has another 20%, you can put the second server in another subnet (possibly at another location) and if your DHCP server goes down, the relay agent will still give the clients an IP off of the second server.
How this works is that the local DHCP server will answer the request for an IP first, and the client will then ACK the request. The relayed request *SHOULD* take longer then the local subnet request, and when it sends the IP to the client, the client should respond with a NACK. So you will not use up two IP's for one machine, but still be covered in the event of a server failure.
Jay
by k4k
by dkolas
by k4k
by k4k
This discussion is closed to new replies.
Sort by Trending