How to apply a Route-Map to a NAT connection

Question

I am following a tutorial about route-maps on youtube. https://www.youtube.com/watch?v=2YlsXcT1Pjw

There is this topology.

The default route on R1 points to 10.10.10.2

In the video he just shows how to configure the route-map itself but i guess to reach the "internet" over the default route he has configured NAT (PAT). So if i configure NAT i map the inside ip to the outside interface. As i am using 10.10.10.2 as gateway i translate the internal ip to 10.10.10.1. But when i apply the route-map to route the traffic out to ISP-B it has to have a source address of R1 interface gi0/2 which would be 20.20.20.1, am i wrong? How do i do that?

Answer 1

You can create 2 route-map entries to cover both cases. Cisco provides this simple example for a kind of round robin style load balancing to use both connections at once. Traffic would be sent via one connection and then the other, alternating with each session.

https://www.cisco.com/c/en/us/support/docs/ip/network-address-translation-nat/100658-ios-nat-load-balancing-2isp.html

It's not a very practical implementation for real life since it may cause issues with browser sessions that are IP based instead of cookie based, or with UDP traffic, etc.

This example provides an improved design via the same features but with the addition of SLA implementation to monitor and fail between the 2 ISP connections based on their ability to ping a remote host.

https://danjmcintyre.com/2020/02/13/cisco-router-dual-wan-uplinks-with-nat/

In real life, you would probably not use IOS routers in this manner, you would use a firewall with built in dual WAN capability which may include failover and smart load balancing based on source/destination address, protocols used, traffic load, etc. Or you might use an SD-WAN platform that would have even more control over traffic management. If you used a large enterprise IOS/IOS-XE type Cisco router, you would likely use BGP to manage traffic over the 2 internet services instead of simple failover or round robin balancing.

Answer 2

ip nat inside source route-map cablemodem interface GigabitEthernet0/1 overload
!
ip access-list extended PROTO-41
 permit 41 any any
!
route-map cablemodem deny 3
 match ip address PROTO-41
!
route-map cablemodem permit 10
 match interface GigabitEthernet0/1

For a Real World™ example. It's as simple as using a route-map as a qualifier (identifier/match) for the traffic that should apply to a NAT rule. A route-map can be far more detailed than a single ACL. (non-RM NAT rule) In my case, I deny ("do not NAT") protocol 41 (tunnel) traffic, and then match (permit) the outgoing interface. In effect, if routing put the traffic on G0/1, this NAT rule applies. (there are other rules for other interfaces. eg. the backup DSL interface.)

Remember, routing happens first on IOS routers. To push specific traffic to specific interfaces requires tweaks to routing, or policy based routing ("PBR") which also uses route-maps.