"Will" <westes-(E-Mail Removed)> wrote in message
news:(E-Mail Removed) ...
> After I created the superscope it works fine, and as a matter of fact
> Windows is binding the DHCP scope to the specific ethernet adapter that
> binds to the class C addresses in the scope. It appears that this
binding
> is done for you automatically however and doesn't require me to set
anything
> explicitly.
>
> As a general comment, if I can speak the requirement as a sentence that
has
> meaning, then by definition it isn't meaningless.
No. People say things all the time that aren't accurate. People most often
speak according to what they are thinking and most problems stem from
people's incorrect thinking which is reflected in what they say when they
describe things. So I try to adjust, or steer, people's thinking into the
correct direction, because if peoples's thinking is correct and pointed in
the right direction the problems they have will most often take care of
themselves because the solution become obvious.
> I think you are being
> too quick to dismiss things you don't like as meaningless.
I am saying that in the context of how the running DHCP service "views"
things.
> If each DHCP scope is properly associated with one Class C network, how is
> it meaningless to want a feature that binds the DHCP scope to one and only
> one ethernet card / Class C?
DHCP scopes are not "bound" to anything at all. The DHCP Service recieves a
DHCP Request,..it checks the request to see what IP Segment the request was
initiated from and makes a DHCP Offer using an IP config that matches the
originating segment. There is no "binding". You can have a Query from one
segment be recieved and properly answered when the Nic it was received on
doens't even match the Segment of the request. That is why you can have one
DHCP server with a single nic and single IP config still supply addressing
for multple segments. The only thing that ever gets "bound" is the running
DHCP Service itself,...not the Scopes within it.
> And, to my great surprise, it already works the way I wanted but Microsoft
> is doing it automatically. The superscope doesn't solve the problem if
you
> try to multinet *a single ethernet card*.
> Microsoft's documentation says it is a requirement to have scopes that
> support different networks be on different ethernet cards/ports.
It isn't that simple,...and the exact scenario context matters. I have 9
IP segments,..one single Nic in the DHCP servers,...no Super Scopes,...
these are 2 DHCP server configured as "redundant servers" and it all works
flawlessly and I don't have to do anything "special" nor jump through any
"hoops".
> To make it work, you need to restrict *all* of the IPs in one of the two
> Class C networks in its DHCP scope, then use just reservations to grab IPs
> in that DHCP scope. All the assignments to machines without
reservations
> then go to the other scope, and it appears to work with quite good
> performance.
Never heard of having to do any of that. It is amazing how people can take
something that works in such a simple way and turn it into such a complex
thing, I see this happen over and over in these groups. Get rid of the
Multi-Net by either reconfiguring the physical cabling or do it via VLANing
and life will be better, the sky will be bluer, the clouds will be puffier,
the birds will be chirpier, and all will be well.
--
Phillip Windell [MCP, MVP, CCNA]
www.wandtv.com
Similar Threads
Linear Mode
