In all recent versions of CVSNT the lockserver is the primary means of handling file locking. There should normally only be one lockserver, which may be shared by multiple repositories. Once running it should require little or no maintenence.
The lockserver provides file-level locking for the server, which allows much greater concurrency than previous versions of CVS. It also provides checkout atomicity which ensures that you always get a coherent view of the repository. The previous method of locking using directory locks on the filesystem is now depreciated and should not be used as it does not have these advantages.
Setting up the lockserver under Windows is handled by the setup program and happens automatically. Under Unix you need to arrange to run the cvslockd on startup - this varies between versions.