Robert
Depends on your ISPs really.
Are these connections using different ISPs ? If so the issue is if you use ISP1 addressing and the ISP1 link fails so you go out via ISP2 but the source IP is still an address from the ISP1 range then traffic will be routed back via ISP1 to the failed link.
So if it is two ISPs in order for this to work you would need each ISP to advertise out the others blocks as well in case of a failure. They may not want to do this.
The other option is to simply use ISP1 addressing on ISP1's link and ISP2 addressing on ISP2's link. For outbound traffic this is relatively easy. You simply need to use route maps with your NAT statements so that you not only match on the source IPs ie. your internal clients but also match the outgoing interface as well so it pick an address from the right range.
If you are hosting services that are accessed from the internet if both address blocks are not advertised out by both ISPs it can become tricky. If you use an ISP1 address then if it ISP1 fails and ISP2 is not advertising out that block then that server is unreachable. You could use two addresses one from each ISP to do the NAT and then have two DNS entries for it so DNS does round robin. But this would mean -
1) both links would be used for inbound traffic to this server
2) if a link fails DNS has no way of knowing this so for half the lookups it still hands out the failed link IP address
So basically internal clients to internet is relatively easy even if the ISPs are not advertising each others blocks. Incoming to servers you are hosting is a lot more difficult and needs careful planning and talking to the ISPs to see what they can do in terms of advertisement of addresses.
Of course if both connections are the same ISP most of the above does not apply.
Jon
