let's say we have 30ish routing instances that may or may not need to be extended to 30ish remote sites. currently that is done via mpls. (or one gre tunnel per instances...)
I'm proposing doing a split tunnel at the remote side, an ipsec tunnel back to central, and doing pbf on the same vsys where the ipsec tunnels land. it would require one pbf rule per routing instance we want to extend (and each instance gets its own /16 to keep the rules simple). the goal is that new remote site deployments only require one ipsec tunnel, one bgp peer, and then layer3 interfaces for any routing instances required. I want this dead simple for folks to deploy.