3. Comparison of Client/Server Model Implementation

3.4 Comparison of Implementation of Client/Server Model's
Studying the two implementations reveals a number of issues, both positive and negative, about each mechanism.

The basic application architectures differ significantly. The socket architecture is a very simple client to server topology. This is in stark contrast to the RMI architecture, which requires the use of an interface to acknowledge the remote services.

As a result of this the actual implementation of the connection is simpler using the socket mechanism which only requires the client to create a socket instance and initialise the connection by specifying the port to use to communicate. However the simplicity of creating the connection is matched only by the complexities faced when communicating through the connection. As we know sockets communicate by transmitting streams of data between client and server. Thus all data needs to be processed into a transmittable form (primitive data type) explicitly at by the sender before data is transmitted. The receiver of the data stream must process the data from its transmitted form into form that allows it to interpret the data.

Contrast this with the RMI implementation where the process of registering the service with the rmiregistry may be more draw out but than simply establishing a direct connection but the benefits this provides is evident in the ease with which remote services are accessed. Remote services can be accessed as if they were local to the client. More importantly, because RMI supports the passing of objects and primitives no processing is required to data during remote transmissions.

Overall, the greater sophistication of the RMI mechanism makes it a far more practical solution for the above application. However the RMI mechanism is only a viable solution in java environments while the socket mechanism could be applied to any language environment.

Next Page