WAN Addressing Options with Routing and Remote Access (171185)

MORE INFORMATION

RRAS requires that a pool of addresses be configured even if no routers or RAS clients will be dialing in. The address pool is used when the router is dialing out or receiving calls. The default option is to use DHCP to assign remote TCP/IP client addresses. If you do not have a DHCP server available, be sure to configure a static pool of addresses. Addresses for demand-dial router links come from the same address pool as addresses for dial-in RAS clients. After the first router or RAS client dials in, the route for the static address pool is added to the router's route table.

Using DHCP Assigned Addresses

When DHCP is used to assign addresses to dial-in routers and RAS clients, the subnet route will be added to the server's route table unless it already exists. When using DHCP superscopes, the subnet route may not already exist because the server can be on a different segment from the RAS pool. This route is advertised, but is not used by the router itself.

NOTE: Because the DHCP route might have the same subnet mask as the local LAN, it would not be correct to route packets for the entire subnet out the RAS interface. A route such as this would prevent RAS clients from reaching the local LAN.

Using a Static Address Pool

A network address and a subnet mask define the static address pool. For configuration instructions, refer to the online help. The subnet used by one side of a demand-dial link must not match the subnet used on the other side. If both routers use the same address range and subnet mask for the RAS address pool, duplicate addresses will be assigned to clients from these RAS pools.

"On Subnet" Address Pool

An "on subnet" address pool uses an address pool that is a subset of a local LAN subnet. In this case, other routers and clients already know how to reach this LAN subnet. The Windows NT router will provide proxy ARP to get the packet to the final destination. Clients and other routers will not realize that these IP addresses are not actually on the LAN. This option is simple in that it does not require any additional routes to reach the connected dial-in RAS clients. However, this option may be limiting depending on your subnetting scheme and the number of dial-in clients you need to support.

Using the "on subnet" address pool, the subnet mask of the LAN will not change or be affected. If you were using the class C address 192.168.1.0 with the mask 255.255.255.0 on the LAN, you could reserve a pool in this same address range for the RAS pool. An address pool defined by the subnet address 192.168.1.248 using a subnet mask 255.255.255.248 creates a pool from 248-255. The addresses 248 and 255 are not used and 249 is used internally by the router. This leaves you with support for a combination of 5 RAS clients or demand-dial links. The proxy functionality of the router will allow this apparent over-lapping of subnet ranges. However, all addresses still need to be unique.

NOTE: Using DHCP (without superscopes) for the address pool will always use "on subnet" addressing and the same subnet mask as the LAN.

"Off Subnet" Address Pool

"Off subnet" addressing uses a unique subnet for the address pool. In this scenario, the other routers need to know how to reach this subnet. This can be done using static routes or a routing protocol such as Routing Information Protocol (RIP) or Open Shortest Path First (OSPF). An "off subnet" address pool could provide more addresses if the LAN subnet does not have enough addresses available.

If your "off subnet" WAN addresses are from a different network ID, RIP will summarize routes into a single network route. Individual subnet routes will not be advertised out an interface that is not on the same network ID as the routes being advertised.