Chapter 7. Enterprise JavaBeans

So, you may ask, how are EJB[1] components different from regular distributed objects built using RMI or local (nondistributed) components defined using the JavaBeans component model? Well, they aren't, really. An EJB component is an RMI object, in the sense that it's exported as a remote object using RMI. And an EJB component is also a JavaBeans component, since it has properties that can be introspected, and it uses the JavaBeans conventions for defining accessor methods for its properties. An EJB is much more than the sum of these parts, however. The EJB architecture provides a framework in which the enterprise bean developer can easily take advantage of transaction processing, security, persistence, and resource-pooling facilities provided by an EJB environment. These facilities don't come free, of course. You need to understand how they work and what rules your EJB object needs to follow in order to participate in these services.

[1]For the sake of space on the page, strain on your eyes, and my time on the keyboard, I'm going to abbreviate "Enterprise JavaBeans" as "EJB" throughout most of this chapter. I hope you don't mind.

Enterprise JavaBeans are useful in any situation where regular distributed objects are useful. They excel, however, in situations that take advantage of the component nature of EJB objects and the other services that EJB objects can provide with relative ease, such as transaction processing and persistence. A good example is an online banking application. A user sitting at home wants to connect to all her financial accounts, no matter where and with whom they may live, and see them tied together into one convenient interface. The EJB component architecture allows the various financial institutions to export user accounts as different implementations of a common Account interface, just as we would do with other distributed object APIs. But since these remote Account objects are also JavaBeans components, the client-side financial application can introspect on the Account objects to determine specialized public properties that certain accounts may have, so they can be shown to the client along with the common account properties. Also, the Account objects can be made into transactional EJB objects, which allows the client to perform a number of account operations within a single transaction, then either commit them all or roll them back. This can be a critical feature in financial applications, especially if you need to ensure that a supporting transfer can be executed before a withdrawal request is submitted. The transactional support in EJB ensures that if an error occurs during the transfer and an exception is raised, the entire transaction can be rolled back, and the client-side application can inform you of the reason.

The EJB component model insulates applications and beans (for the most part) from the details of the component services included in the specification. A benefit of this separation is the ability to deploy the same enterprise bean under different conditions, as needed by specific applications. The parameters used to control a bean's transactional nature, persistence, resource pooling, and security management are specified in separate deployment descriptors, not embedded in the bean implementation or the client application. So, when a bean is deployed in a distributed application, the properties of the deployment environment (client load levels, database configuration, etc.) can be accounted for and reflected in the settings of the bean's deployment options.

The EJB API is a standard extension to Java, available in the javax.ejb package and its subpackages. You have to explicitly install this extension API in order to write code against the EJB interfaces. You can find the latest version of the API at http://www.javasoft.com/products/ejb/. You should also note that EJB is just a specification for how distributed components should work within the Java environment. In order to actually create and use EJB objects, you need to install an EJB-enabled server.

Note that this chapter provides a basic introduction to Enterprise JavaBeans. For more complete coverage, see Enterprise JavaBeans by Richard Monson-Haefel (O'Reilly).

7.1. A Note on Evolving Standards

The information and code examples in this chapter are based on Version 1.0 of the Enterprise JavaBeans specification, released in March 1998. The code examples have been tested in two different EJB servers for compatibility: Weblogic/BEA's Tengah server Version 3.1.2 and the 0.4 Version of the free reference EJB server provided by the good folks at EJBHome (http://ejbhome.iona.com).

At the time of this writing, Sun has released a public draft of Version 1.1 of the EJB specification and has plans for a 2.0 version, to be released at a later date. The information in this chapter is largely unchanged by the incremental 1.1 update, but some details on the changes in 1.1 are described at the end of the chapter. It's unclear how much of the material in this chapter will be applicable to EJB 2.0.


6.12. Searching a Directory		7.2. EJB Roles

Chapter 7. Enterprise JavaBeans

Contents:

7.1. A Note on Evolving Standards