The Java EE Connector Architecture defines system contracts that enable resource adapter life cycle and thread management.
The Connector Architecture specifies a life cycle management contract that allows an application server to manage the life cycle of a resource adapter. This contract provides a mechanism for the application server to bootstrap a resource adapter instance during the deployment or application server startup. It also provides a means for the application server to notify the resource adapter instance when it is undeployed or when an orderly shutdown of the application server takes place.
The Connector Architecture work management contract ensures that resource adapters use threads in the proper, recommended manner. It also enables an application server to manage threads for resource adapters.
Resource adapters that improperly use threads can jeopardize the entire application server environment. For example, a resource adapter might create too many threads or it might not properly release threads it has created. Poor thread handling inhibits application server shutdown. It also impacts the application server's performance because creating and destroying threads are expensive operations.
The work management contract establishes a means for the application server to pool and reuse threads, similar to pooling and reusing connections. By adhering to this contract, the resource adapter does not have to manage threads itself. Instead, the resource adapter has the application server create and provide needed threads. When the resource adapter is finished with a given thread, it returns the thread to the application server. The application server manages the thread: It can return the thread to a pool and reuse it later, or it can destroy the thread. Handling threads in this manner results in increased application server performance and more efficient use of resources.
In addition to moving thread management to the application server, the Connector Architecture provides a flexible model for a resource adapter that uses threads:
The requesting thread can choose to block (stop its own execution) until the work thread completes.
Or the requesting thread can block while it waits to get the work thread. When the application server provides a work thread, the requesting thread and the work thread execute in parallel.
The resource adapter can opt to submit the work for the thread to a queue. The thread executes the work from the queue at some later point. The resource adapter continues its own execution from the point it submitted the work to the queue, no matter when the thread executes it.
With the latter two approaches, the submitting thread and the work thread may execute simultaneously or independently from each other. For these approaches, the contract specifies a listener mechanism to notify the resource adapter that the thread has completed its operation. The resource adapter can also specify the execution context for the thread, and the work management contract controls the context in which the thread executes.