Just taking a step back Paul, is the existing solution actually working - i.e. does it load balance and does it failover OK? It's possible that you are best leaving it alone, especially if the solution is relatively easy to understand and manage. If not, then can you test any proposed solution in a lab?
Attached is a couple of diagrams. I tested Topology A in a lab and it was implemented successfully. Topology B is a simplified version that remains untested. I'm sorry they are a bit basic, but I don't have time right now to flesh them out.
TA used an IGP between WAN Edge routers and the L3 switches, with redistribution between OSPF and BGP, carefully filtered of course. IP CEF was used to load balance outbound traffic. The ISP circuits were both 100Mbps. WAN Edge routers were 2911s, WAN L3 switches 3560-X series, firewalls were 5520s and the core was 6500s. Inbound load balancing used BGP AS_PATH attribute as described in a previous post. HSRP is running on both the inside and outside to provide FHRP for the ASA active/standby pair.
TB avoids the need for the above descirbed IGP and, in theory, could instead rely heavily on IP SLA pings tied to static routes. I think it's a simpler design to manage. There is a lot of detail in TA that is not obvious until you try to make it work, especially when a similar implementation was split across two sites several Km apart. It was not easy to document and explain, especially when the customer needed all the (dozens) of possible failure scenarios documented.
