Hi VirtualNet, thank you for your feedback around this current behavior and great explanation of your use case and how it can affect the routing in your environment. We already had an enhancement in plan for our April feature release, but I'm working with our development team to see if we can adjust this behavior within the next week as part of a patch. This change would allow users to apply (removing route filter policies that are already applied when BGP is disabled already behaves this way) Route Filter Policies to a connection that is configured with BGP disabled, or where BGP is otherwise in a disabled state. Once this change goes in, you would be able to utilize the existing ability to configured BGP and leave it in a Disabled state, then apply your Route Filter Policies before enabling BGP.
In the very short-term while we work to get a patch rolled out to enhance the existing behavior, there is an approach that may be helpful in working around this behavior: As you create your connections, you can configure the Layer 3 / BGP details on the Equinix side first so that BGP is enabled but will not move into an Established state, allowing you to apply your Route Filter Policies before configuring BGP on the remote peer.
As soon as I have confirmation around the date/time that this patch is being merged to production, I'll be sure to follow up and share these details.
