Monkey in the Middleware

P&S is just one approach to middleware, a category of software that provides a mechanism for connecting different and disparate systems and, increasingly, for supporting truly distributed computing. The following list describes some other middleware approaches.

Remote procedure call (RPC) architectures allow developers to invoke procedures on a server remotely through a procedure call from the client. The server responds to the RPC as if it were invoked locally, simply accepting the client's request and fulfilling it synchronous ly. Programming with RPCs can be complicated because each client must be designed to interface with the appropriate procedure on the server.

Object request brokers (ORBs) define standards for constructing object-oriented distributed applications. The Common Object Request Broker Architecture (CORBA) defines standards for distributed computing over TCP/IP networks using ORBs to pass requests and responses from any client to any server, as long as they are employing CORBA interfaces.

Transaction processing (TP) monitors coordinate and manage distributed updates across multiple databases from a large volume of clients. These monitors manage interactions between clients and servers with a high degree of reliability, maintaining the state of client sessions to provide transactional-quality service, but usually at the price of a complex implementation.

Message-queueing middleware uses the concept of the queue to transmit messages from one system to another. Relying on queues allows these systems to handle the transmission of data asynchronously between systems that may not be on-line at the same time. The message-queue software forwards a message to its destination when the destination is reachable, saving the message for later delivery if the destination is unavailable. Message queueing offers extremely high reliability, but its relatively low performance can make it unsuitable for interactive applications.