The primary key can't be a primitive type--use the primitive wrapper classes, instead. For example, you can use java.lang.Integer as the primary key class, but not int (it has to be a class, not a primitive)

The developer, according to the specification, cannot directly control when passivation occurs. Although for Stateful Session Beans, the container cannot passivate an instance that is inside a transaction. So using transactions can be a a strategy to control passivation.

The ejbPassivate() method is called during passivation, so the developer has control over what to do during this exercise and can implement the require optimized logic.

Some EJB containers, such as BEA WebLogic, provide the ability to tune the container to minimize passivation calls.

Taken from the WebLogic 6.0 DTD -"The passivation-strategy can be either "default" or "transaction". With the default setting the container will attempt to keep a working set of beans in the cache. With the "transaction" setting, the container will passivate the bean after every transaction (or method call for a non-transactional invocation).

Entity Beans actually represents the data in a database. It is not that Entity Beans replaces JDBC API. There are two types of Entity Beans Container Managed and Bean Mananged. In Container Managed Entity Bean - Whenever the instance of the bean is created the container automatically retrieves the data from the DB/Persistance storage and assigns to the object variables in bean for user to manipulate or use them. For this the developer needs to map the fields in the database to the variables in deployment descriptor files (which varies for each vendor).

In the Bean Managed Entity Bean - The developer has to specifically make connection, retrive values, assign them to the objects in the ejbLoad() which will be called by the container when it instatiates a bean object. Similarly in the ejbStore() the container saves the object values back the the persistance storage. ejbLoad and ejbStore are callback methods and can be only invoked by the container. Apart from this, when you use Entity beans you dont need to worry about database transaction handling, database connection pooling etc. which are taken care by the ejb container. But in case of JDBC you have to explicitly do the above features. what suresh told is exactly perfect. ofcourse, this comes under the database transations, but i want to add this. the great thing about the entity beans of container managed, whenever the connection is failed during the transaction processing, the database consistancy is mantained automatically. the container writes the data stored at persistant storage of the entity beans to the database again to provide the database consistancy. where as in jdbc api, we, developers has to do manually.

EJB QL is a Query Language provided for navigation across a network of enterprise beans and dependent objects defined by means of container managed persistence. EJB QL is introduced in the EJB 2.0 specification. The EJB QL query language defines finder methods for entity beans with container managed persistenceand is portable across containers and persistence managers. EJB QL is used for queries of two types of finder methods: Finder methods that are defined in the home interface of an entity bean and which return entity objects. Select methods, which are not exposed to the client, but which are used by the Bean Provider to select persistent values that are maintained by the Persistence Manager or to select entity objects that are related to the entity bean on which the query is defined.

EEJB was originally designed around remote invocation using the Java Remote Method Invocation (RMI) mechanism, and later extended to support to standard CORBA transport for these calls using RMI/IIOP. This design allowed for maximum flexibility in developing applications without consideration for the deployment scenario, and was a strong feature in support of a goal of component reuse in J2EE.

Many developers are using EJBs locally -- that is, some or all of their EJB calls are between beans in a single container.

With this feedback in mind, the EJB 2.0 expert group has created a local interface mechanism. The local interface may be defined for a bean during development, to allow streamlined calls to the bean if a caller is in the same container. This does not involve the overhead involved with RMI like marshalling etc. This facility will thus improve the performance of applications in which co-location is planned.

Local interfaces also provide the foundation for container-managed relationships among entity beans with container-managed persistence.

EIt is important to understand that the calling semantics of local interfaces are different from those of remote interfaces. For example, remote interfaces pass parameters using call-by-value semantics, while local interfaces use call-by-reference.

This means that in order to use local interfaces safely, application developers need to carefully consider potential deployment scenarios up front, then decide which interfaces can be local and which remote, and finally, develop the application code with these choices in mind.

While EJB 2.0 local interfaces are extremely useful in some situations, the long-term costs of these choices, especially when changing requirements and component reuse are taken into account, need to be factored into the design decision.

In case of a stateless session bean it may not matter if we call or not as in both cases nothing is done. The number of beans in cache is managed by the container.

In case of stateful session bean, the bean may be kept in cache till either the session times out, in which case the bean is removed or when there is a requirement for memory in which case the data is cached and the bean is sent to free pool.

In several ways, the dynamic creation and allocation of message-driven bean instances mimics the behavior of stateless session EJB instances, which exist only for the duration of a particular method call. However, message-driven beans are different from stateless session EJBs (and other types of EJBs) in several significant ways:

Message-driven beans process multiple JMS messages asynchronously, rather than processing a serialized sequence of method calls.

Message-driven beans have no home or remote interface, and therefore cannot be directly accessed by internal or external clients. Clients interact with message-driven beans only indirectly, by sending a message to a JMS Queue or Topic.

Note: Only the container directly interacts with a message-driven bean by creating bean instances and passing JMS messages to those instances as necessary.

The Container maintains the entire lifecycle of a message-driven bean; instances cannot be created or removed as a result of client requests or other API calls.

EJBs can be clients of other EJBs. It just works. Use JNDI to locate the Home Interface of the other bean, then acquire an instance reference, and so forth.

EJBContext is an interface that is implemented by the container, and it is also a part of the bean-container contract. Entity beans use a subclass of EJBContext called EntityContext. Session beans use a subclass called SessionContext. These EJBContext objects provide the bean class with information about its container, the client using the bean and the bean itself. They also provide other functions. See the API docs and the spec for more details.

The EJB container maintains an instance pool. The container uses these instances for the EJB Home reference irrespective of the client request. While refering the EJB Object classes the container creates a separate instance for each client request. The instance pool maintainence is up to the implementation of the container. If the container provides one, it is available otherwise it is not mandatory for the provider to implement it. Having said that, yes most of the container providers implement the pooling functionality to increase the performance of the application server. The way it is implemented is again up to the implementer.