The basic idea behind procedure oriented middleware is that as far as the client process is concerned, there is no logical difference between calling a local procedure or calling a remote procedure.

Remote Procedure Calls (Top)

A remote procedure call uses direct, connection-oriented, synchronous communication to allow a client process to call a remote procedure.

Enabling a Call

The following steps are required to enable a remote procedure call:

1. Code the procedure name and parameters using the syntax of the Interface Definition Language (IDL) provided by the middleware.

2. Use the IDL to create the source code for a client stub and a server skeleton.

3. Compile and link the stub into the client process.

4. Compile and link the skeleton into the server process.

Making a Call

To make a remote procedure call:

1. The client process calls the client stub as if it was a local procedure.

2. The client stub converts the parameters into a string of bits (called marshalling) and sends the bits over the network to the server skeleton.

3. The server skeleton converts the bits back to parameters (called demarshalling) and calls the server procedure.

4. The server skeleton converts the procedure response into a string of bits and sends then over the network to the client stub.

5. The client stub converts the bits to the response and returns it to the calling procedure.

Mechanisms

The best known RPC mechanisms are:

• Distributed Computing Environment (DCE) from the Open Software Foundation (OSF).

• Open Network Computing (ONC) from SUN.