Considering the fact that some query handling and security needs to be handled for each server.
Yes, this is a common deployment pattern for FDB, and is the recommended way to deploy the document layer currently.
The document layer seems to couple the layer client with the application server in its sidecar model while the cluster is kept separate.
However, I would like to know whether the FDB database server process and the FDB client(another process) can be deployed on the same machine instead of maintaining it as a separate cluster?
Yep, I clearly misunderstood you there. Sorry!
That is also a reasonable deployment pattern, which was described by WaveFront for their system at the FDB Summit.
You can also use this to implement a primitive form of predicate pushdown in your layer, as a read from a storage process colocated on the same machine only requires local networking. Building a router to send requests to the right colocated layer process is left as an exercise for the reader 