Subject: RE: MCSE TCP/IP question
Date sent: Sun, 20 Jul 1997 20:17:37 -0500

I am taking the TCP/IP for NT 4.0 soon, but am still a little shaky on some things. Here's the question...

Your company has a great demand for Internet access, but only one class C address. You install DHCP servers on each of your 6 subnets to assign IP addresses dynamically. After creating DHCP scope ranges of 30 addresses for each subnet, you want to provide for redundancy, should a DHCP server go down, by creating a second scope on each server that contains 5 addresses from it's neighboring subnet, as shown by the breakout of the 4th octet:

SubnetA 33-62 65-69

SubnetB 65-94 97-101

SubnetC 97-126 129-133

SubnetD 129-158 161-165

SubnetE 161-190 193-197

SubnetF 193-222 33-37

How well does this solution address the problem?

The practice test I got this from indicated this meets the requirements but is not a desirable solution, because the addresses in the second scopes would cause duplicate IP's if one of the servers goes down.

Here's where I don't understand. I thought as long as your IP addresses are unique to the subnet, you could have duplicates in other subnets. Example: SubnetA

123.47.89.1, and the host address in SubnetB. I feel confused. Help!

The following is an answer to this question >>>>>>>

Steve,

The answer given in the test is correct, because it is possible for a DHCP server to assign IP address to computers on a different subnet, if the routers pass DPCH broadcasts, and the servers are configured with proper scopes.

In this scenario each subnet can have a maximum of 32 host addresses. The allowable subnet ranges would be 32-63, 64-95, 96-127, 128-159, 160-191, 192-223, and the subnet mask would be 255.255.255.224.

The first column of numbers in the table above are the subnet addresses that a DHCP server will assign to computers on its own subnet (assuming that it answers the DHCPDISCOVER message for the computers on its own subnet before another DHCP server across a router). The second column of numbers correspond to subnet addresses for an adjacent subnet, and would only be assigned to computers on that subnet, and only if the DHCP server on that subnet failed to answer the DHCPDISCOVER message first.

The proper way to configure the servers (at least per Microsoft) is to assign 75% of the local subnet addresses to the local DHCP server, and 25% to an a server existing on another subnet. Using the same numbers above this would be a better solution:

SubnetA 38-62 65-69

SubnetB 70-94 97-101

SubnetC 102-126 129-133

SubnetD 134-158 161-165

SubnetE 166-190 193-197

SubnetF 197-222 33-37

With this configuration there would be no chance that two DHCP servers would attempt to assign the same IP address on the network.

-Gil