Question on Transcender Question :-)

I have a Win2K3 AD design transcender question in which you have two LANs
connected by a WAN link. A requirement of the DHCP implementation is to
provide redundancy in case of the failure of one server.

Two possibile answers are:

b) - Install a DHCP server in each office. On each server configure a single
scope and an exclusion range.
d) - Install a DHCP server in each office. On each server configure two
scopes, for each scope configure an exclusion range.

The recommended answer is (d). I do not understand this answer.

Let's say that I have two subnets:

LAN1 - 10.0.15.0/24
LAN2 - 10.0.16.0/24

connected by a WAN link.

The DHCP server on LAN1 is set up to service DHCP requests from LAN2 as well
as LAN1. Fine. But how will it know not to assign LAN2 IP addresses to local
computers. If it does that the local computer will belong to the wrong
subnet and will not be able to access any local resources?

TIA

eddiec

 

Any relation to EddiePok/EddieRox?

 

This is the age-old eighty-twenty rule in action

two subnets, two dhcp servers.
Server on Subnet A
subnet A 80% of address range
subnet B 20% of address range

Server on Subnet B
subnet B 80% of address range
subnet A 20% of address range

Which means that if one server is down the other can temporarily issue
new leases to give you time to recover - in reality this means a 4
hour window to an 8 hour window to get things fixed with minimised
impact on your environment.

Your question on what addresses get assigned to which subnet is a good
one. If your router is bootp/rfc1542 compliant then the discover
broadcast from the other subnet will arrive at the dhcp server tagged
with that fact in its header and the dhcp server will issue an address
based on the route it took to get to the dhcp server.

This means that server on subnet B *can* issue addresses for subnet A
because the DORA has the correct tag information from the Subnet A
interface of the router. Clever eh!!?


================================
My Hero:
http://www.theregister.co.uk/content/30/index.html
(remove vroomfondle to email me)
================================

 

Nice Info

 

Look at it from the other direction...why would you want local clients to
get IP assignments over the links, from a distant source?

It is much more logical to have local assignment of IP addresses. Cuts down
on delay, overhead, frees up bandwidth...

I can't believe that Transcender did not give a very lucid explanation to
the question. Are you real sure you read it and looked up the references?
You can't just use the simulation as a sole training aid. If you are really
confused, Transcender does offer a help line/email contact.

PS. Cross posting is usually discouraged. Instead of generating one thread,
you are generating x times the number of post, plus those replies. It
appears that Usenet bandwidth (traffic conservation) is a thing of the past,
as are many other courtesies

 

Look at it from the other direction...why would you want local clients to

In general, you're right: IP addresses should be obtained locally as much as
possible. The answer however lays in the question: you have to provide
redundancy. In other words: you will set up the DHCP servers as described in
answer D. As a result, computers will get their IP addresses locally, until
that server fails or runs out of addresses.

Best regards,
Pieter

P.s.: one of the courtesies you are talking about is not replying at the top
of a posting, but at the bottom. Another one is quoting: deleting the parts
of the message you are responding to that are not relevant.

 

I had to think about that. Since a large number of people do "top post" as
you call it...

I don't think you are right. If you read the original post (and subsequent
replies) top posting allows you to read the reply without going through the
entire history in each and every message. What you may be thinking about is
"snipping" ... truncating the original message and/or previous replies.

I hope you are not advocating cross posting?