Under the classic model, remote client requests are processed individually by the database server. Using the threaded model, remote client requests are processed concurrently. A threaded server can spawn and manage client threads up to a maximum of the value specified by the -Ma parameter.

Concurrent processing of remote client requests improves database performance and provides for horizontal scalability. Concurrent processing also ensures continuous availability as remote client process kills will not crash the database.

Both the threaded database server and the server-side join startup parameters are turned on by default.

However, server-side joins are only supported on multi-threaded database servers that require at least the Enterprise license or, on a higher level, the Advanced Enterprise RDBMS license. Without the Enterprise database license, the server-side join feature is disabled by default.