Discovery can be slow in environments where DCs are offline or unreachable.
In a test environment with about 150 DCs scattered around the globe, with about a dozen of those DCs being unreachable and causing connection timeouts, discovery takes around 20 minutes.
I feel this could be improved by breaking up the total number of DCs to scan into chunks and using multiple worker threads to break up the work.
But in addition, connection timeouts need to be adjusted if possible so we don't wait so long for a timeout if the DC is unreachable.