2. RMI versus Remote Process Calls (RPC)

2.1 Remote Method Invocation (RMI)

RMI was developed by Sun to provide remote communication solely between Java objects. The idea behind this is that a Java dedicated mechanism would be able to take advantage of all of Java's key object and security features otherwise ignored by more general communication mechanisms.
Thus RMI is a mechanism that allows client java objects to connect to remote java objects and use them as if they were local to that client. A look at the RMI architecture reveals how this is achieved.
Firstly, the RMI architecture separates the code that defines the behaviour from that which implements the behaviour enabling them to run on separate Java Virtual Machine's (JVM). It should be noted that a Java interface contains no executable code. Instead RMI supports two classes that implement the same interface. The first class, run on the server, is the implementation of the behaviour and the second class, run on the client acts as a proxy for the remote service. Figure 1 illustrates this structure.

Figure 1. RMI implementation structure [2]

To allow the client and server objects to forward method calls between one another RMI employs a proxy pattern, known as a Stub and Skeleton. Figure 2 illustrates the stub and skeleton role.

Figure 2. Role of the Stub and Skeleton in RMI [2]

When a method is 'called on the remote object' the call is actually made on the local proxy object or stub object. The stub communicates with a skeleton object in same space as the remote object being accessed. This communication is performed transparent to the client and it is the function of the stub-skeleton pair to hide the transport and session layers, in particular note the conversion of method parameters (including objects) to and from byte streams.

All the required classes for RMI implementation are included in the JDK development kit (java.rmi.*).

Next Page