Mongodb instances and oplog

I would like to ask a couple of things about mongodb and replica sets:

I have different meteor applications running on my private server; each one of them uses a different meteor database (and does not need to access other databases); right now only one mongod instance is running on the server, serving all applications; should I run different mongod instances (one for every application)?

I would like to create a replica-set with oplog; my question is: this is only useful for performance when there are a lot of requests (coming from many different clients) or will improve performance also in every single request? (The different members of the replica set will cooperate to find faster the documents requested by a single client?)