Thanks for the response. My end goal is to have PDAs in the field dial-in, connect to this router, and synch info with a server on my internal network. I use a /21 network so don't I have to declare that in the loop back? Then provide a pool of IPs that reside in that subnet so they can communicate with my internal network?

I'm not sure what you mean by "if you use only one asynchronous inerface, you would need the loopback neither."

Yes, you can declare the ip pool to be in the same internal lan, however to make things nicer I would use a different subnet.

Sometime people especially ISP have multiple group-async interfaces, so to save on addresses, they make them all unnumbered off the same loopback. But if you have a single group-async, is not necessary to create a loopback only for the purpose, as it works anyway no matter which address you give it it. However, the cleaner way is a subnet that comprises the ip pool.

OK, I appreciate the help but that's a lot of information there I'm not familiar with and I'm a bit lost. So do I need to declare the loopback as 'ip unnumbered' or declare it with a static ip in the sunbnet that I'm going to assign the ip pool in?

I need the dial-in users to be able to talk to my internal server for synch processing so I'm going to make the ip pool within my internal subnet.

Internal - 192.168.0.0 255.255.248.0

Proposed Dial-in IP Pool - 192.168.7.201 - 192.168.7.224

Loopback - IP Unnumbered or 192.168.7.200 255.255.248.0? Does it matter?

Also, I'm confused about the group-async interface and why it has it's own loopback interface. I have never set something like this up before and am trying to follow step by step with the document http://www.cisco.com/en/US/docs/ios/12_4t/12_4t11/htpvdm2.html

in the 'Tasks Common to all configurations' section. It's a bit overwhelming as I don't have experience in this and am be pushed hard to get it working. All help is greatly appreciated.

Another problem...I tried to assign the loopback 0 interface with an IP on my local subnet but I got an error back stating "% 192.168.0.0 overlaps with FastEthernet0/0" so I guess I have to give it an address on a different subnet.

If I do that and all dial-in users are assigned IPs with a /32 how does it know to route that traffic into my interal LAN?

I am responding to both your question here:

Loopback 0 needs to have a static address which we will use for the group-async interface (using "ip unnumbered loopback 0")

This static address can be from the same pool subnet or a totally different subnet however like any router requirement - 2 interfaces cannot be from the same subnet (so your Ethernet and loopback are overlapping).

To make it simple, give the loopback any ip address, say 1.1.1.1 255.255.255.255 and have the group-asynch unnumbered to loop 0.

As said earlier all dial in users will be assigned a host route (/32) from the address pool, any packet from the dialin client will ve forwarded to the router (def route) which will further forward it to the internal LAN (for the router will appear as directly connected).

Let me know if you have any further questions.

Thanks, Mak

That's much clearer, thank you.

Do I need to have the default ip pool that dial-in users are going to pull from in the same subnet as the loopback?