Custom essays on WWW, CORBA and Java technologies in the construction of distributed object systems

Construction of interaction environment is one of the most difficult steps in developing an information system. As practice shows, the creation of a unique environment of interaction of objects by information system developers leads, on one side, to a sharp increase of costs for the information system construction, and on the other side, to the incompleteness (defectiveness) of the solution.
Research of information system development projects allow us to conclude that in order to avoid unnecessary costs for the development of a unique information environment, it is necessary to use existing software products that relate to the level of middleware and create much-needed interaction environments. However, not all the middleware products can be used as an environment for interaction between objects of a large information system, because one of the main requirements for a large information system is the use of software products and technologies that satisfy international and industry standards in the field of open information systems. In this regard, today, the products supporting CORBA standard are increasingly considered as candidates for realizing high-level interaction environments (Tanenbaum & Steen, 2002).
In this paper we try to justify theoretically that currently Java and CORBA technologies are the best suited for the integration of WWW technology-based systems with large information systems. That is, the use of these technologies allows raising the WWW technology to a new level – the level of distributed systems.

Distributed object technology
Today, many companies and manufacturers are absorbed by the breakthrough in the field of Internet-based distributed client/server systems based on the model of distributed objects.
A system based on distributed object technology consists of a set of interacting components (objects) which are usually scattered over a network and executed separately from each other. Using the distributed objects technology allows taking full advantage of object-oriented approach: reducing development time (stand-alone development), reducing errors, re-using software components, facilitating future changes in the system. Another important advantage of such systems is the possibility to construct so-called light clients suitable for fast downloading via the network (Internet) and launching on the client’s computer (such applications include applets or Active-X components).
Currently, there are three different technologies that support the concept of distributed object systems: RMI, CORBA and DCOM.
Architecture of RMI (Remote Method Invocation) which is integrated with JDK1.1 is a JavaSoft product and implements a distributed computing model. RMI allows client and server applications to invoke methods of clients/servers running on the Java Virtual Machine. Although RMI is considered lightweight and less powerful than CORBA and DCOM, it has several unique features, such as a distributed automated object management and the ability to send objects from machine to machine. Client Stub and Server Stub are generated from a common interface, but the difference between them is that the client stub is used simply to connect to the RMI Registry, and the server stub is used to communicate directly with the server functions (Pic. 1) (Bonsangue & Johnsen, 2007).

Pic. 1: RMI Model
CORBA Technology (Common Object Request Broker Architecture), developed by the OMG (Object Managment Group) in 1990, allows activating methods in objects located anywhere in the network, as if they were local objects. Dynamic Invocation Interface (DII) allows the client to find servers and invoke their methods during system operation. IDL Stubs defines how the client makes a server request. ORB Interface presents services common for both client and server. IDL Skeleton provides static interfaces to the objects of certain type. Dynamic Skeleton Interface presents common interfaces for objects, irrespective of their type, which were not defined in IDL Skeleton. Object Adapter provides communication interaction between the object and the ORB (Pic.2) (Tanenbaum & Steen, 2002).

Pic. 2: CORBA ORB
DCOM Technology (Distributed Component Object Model) was developed by Microsoft as a solution for distributed systems in 1996. DCOM is now the main competitor to CORBA, although it is no longer controlled by Microsoft, but by the TOG (The Open Group), similar to OMG. In short, DCOM is an extension of COM architecture to the level of network applications (Pic.3) (Tanenbaum & Steen, 2002).

Pic.3: DCOM architecture

The benefits of CORBA over RMI and DCOM technology
Each of the three considered technologies has its own unique characteristics, which largely characterize the possibility or impossibility of its application for the solution of a concrete task.
The advantages of DCOM include linguistic independence; dynamic / static request; dynamic detection of objects; scalability; open standards (TOG control). Its disadvantages are complexity of implementation; dependence on the platform; absence of naming through URL; absence of security checks at the level of implementation of ActiveX components.
DCOM is only a particular solution to the problem of distributed object systems. It is well applied for Microsoft-oriented environments, but once the system is to work with the architecture different from Windows, DCOM ceases to be an optimal solution. Of course, this situation may change, as Microsoft seeks to move DCOM to other platforms. For example, Software AG has released a DCOM version for Solaris UNIX and plans to release versions for other versions of UNIX. Another big concern is the lack of security in the performance of ActiveX component which can lead to serious consequences.
The advantages of RMI include quick and easy creation; Java-optimization; dynamic loading of component adapters; opportunity to transfer objects by value; built-in security. Its disadvantages are support for only one language (Java); presence of its own (not IIOP-compatible) protocol interaction; difficulties in integrating with existing applications; and poor scalability.
Due to its easily applied Java-model, RMI is the simplest and fastest way to build distributed systems. RMI is a good choice for creating RAD-components and small applications in Java. However, RMI isn’t such a powerful technology as DCOM or CORBA. In particular, RMI uses its own, not CORBA/IIOP-compliant JRMP protocol and can communicate only with other Java objects. Single language support makes it impossible to interact with non-Java written objects. Thus, the role of RMI in creating large, scalable, industrial systems is reduced (Lin & Veeravalli, 2003).
The advantages of using CORBA technology are platform independence; language independence; dynamic requests; dynamic detection of objects; scalability; presence of CORBA-services; and broad industrial support. Its disadvantages include the absence of parameter transmission by value; absence of dynamic loading of component adapters; and absence of naming through URL.
However, CORBA is the most efficient, modern, suitable technology for large projects, because, although both technologies – CORBA and DCOM – are extremely similar in functionality and features (multi-language support, dynamic call, scalability, etc.), DCOM lacks a critical element – multiplatform support. In addition, while OMG now includes more than 700 members (manufacturers of software products, computers, and telecommunications systems; application developers and end users), and virtually any specification developed by this consortium effectively becomes the standard. Another advantage of CORBA technology is that the range of producers supporting this technology is much broader than DCOM’s one. Thus, CORBA is a technology, which is fully intended for industrial, open, distributed object systems (Bonsangue & Johnsen, 2007).