Gateways and Kernel Provisioners

Hello! I’ve been using Enterprise-Gateway for a bit now and have some questions about it’s future position with the inclusion of Kernel Provisioners in jupyter_client 7.0. As I’ve been operating Enterprise Gateway a desire to have multiple infrastructures, i.e. some kernels local, some in K8s, some in Yarn, all within one JupyterLab, has arisen. I see multi-gateway support on client-side is on the wish list of features for Enterprise Gateway, but as I’m familiarizing myself with Kernel Provisioners I’m wondering if that wish list item is out of date?

Please correct me if I’m wrong, but my sense is that Kernel Provisioners solves this problem. See my attached diagram for my mental model of JupyterLab + Kernel Provisioners + Enterprise Gateway

I know that Enterprise Gateway doesn’t currently support Provisioners, but otherwise is this idea roughly correct? And where would Kernel Gateway fit into this picture? My sense is that it’s a 1->1 mapping between kernel and gateway, whereas Enterprise Gateway is a 1->n mapping. Enterprise Gateway seems like the easier setup networking-wise, for interacting w/ kernels from outside a given cluster.

Thanks!