Documentation on best practices configuration for large numbers of workers

I don't *think* this is an existing issue or anywhere in the docs, but my apologies if it is already mentioned somewhere. I occasionally run jobs on 1500+ worker clusters (using dask-gateway with the Kubernetes backend) from an interactive jupyterlab interface. When I do this, the notebook response time slows down considerably and I'm assuming this has to do with some extra communication thats occurring between client and scheduler due to the extra worker(?) If that's true, I can see a number of configuration parameters that might improve this situation by slowing down communication frequency, but I'm not exactly positive which ones to alter. I'm wondering if it could be useful to have a "best practices for creating clusters with many workers" section of the docs which details some tips and tricks for workflows on large clusters. It's also possible there is no easy answer to this question, in which case feel free to close this feature request.