One of the key themes of the Spring Framework is inversion of control. In its broadest sense, this means that the framework handles responsibilities on behalf of the components that are managed within its context. The components themselves are simplified since they are relieved of those responsibilities. For example, dependency injection relieves the components of the responsibility of locating or creating their dependencies. Likewise, aspect-oriented programming relieves business components of generic cross-cutting concerns by modularizing them into reusable aspects. In each case, the end result is a system that is easier to test, understand, maintain, and extend. Furthermore, the Spring framework and portfolio provide a comprehensive programming model for building enterprise applications. Developers benefit from the consistency of this model and especially the fact that it is based upon well-established best practices such as programming to interfaces and favoring composition over inheritance. Spring's simplified abstractions and powerful support libraries boost developer productivity while simultaneously increasing the level of testability and portability. Spring Integration is a new member of the Spring portfolio motivated by these same goals and principles. It extends the Spring programming model into the messaging domain and builds upon Spring's existing enterprise integration support to provide an even higher level of abstraction. It supports message-driven architectures where inversion of control applies to runtime concerns, such as when certain business logic should execute and where the response should be sent. It supports routing and transformation of messages so that different transports and different data formats can be integrated without impacting testability. In other words, the messaging and integration concerns are handled by the framework, so business components are further isolated from the infrastructure and developers are relieved of complex integration responsibilities. As an extension of the Spring programming model, Spring Integration provides a wide variety of configuration options including annotations, XML with namespace support, XML with generic "bean" elements, and of course direct usage of the underlying API. That API is based upon well-defined strategy interfaces and non-invasive, delegating adapters. Spring Integration's design is inspired by the recognition of a strong affinity between common patterns within Spring and the well-known Enterprise Integration Patterns as described in the book of the same name by Gregor Hohpe and Bobby Woolf (Addison Wesley, 2003). Developers who have read that book should be immediately comfortable with the Spring Integration concepts and terminology.

Spring Integration is motivated by the following goals: Provide a simple model for implementing complex enterprise integration solutions. Facilitate asynchronous, message-driven behavior within a Spring-based application. Promote intuitive, incremental adoption for existing Spring users. Spring Integration is guided by the following principles: Components should be loosely coupled for modularity and testability. The framework should enforce separation of concerns between business logic and integration logic. Extension points should be abstract in nature but within well-defined boundaries to promote reuse and portability.

From the vertical perspective, a layered architecture facilitates separation of concerns, and interface-based contracts between layers promote loose coupling. Spring-based applications are typically designed this way, and the Spring framework and portfolio provide a strong foundation for following this best practice for the full-stack of an enterprise application. Message-driven architectures add a horizontal perspective, yet these same goals are still relevant. Just as "layered architecture" is an extremely generic and abstract paradigm, messaging systems typically follow the similarly abstract "pipes-and-filters" model. The "filters" represent any component that is capable of producing and/or consuming messages, and the "pipes" transport the messages between filters so that the components themselves remain loosely-coupled. It is important to note that these two high-level paradigms are not mutually exclusive. The underlying messaging infrastructure that supports the "pipes" should still be encapsulated in a layer whose contracts are defined as interfaces. Likewise, the "filters" themselves would typically be managed within a layer that is logically above the application's service layer, interacting with those services through interfaces much in the same way that a web-tier would.

In Spring Integration, a Message is a generic wrapper for any Java object combined with metadata used by the framework while handling that object. It consists of a payload and header and has a unique identifier. The payload can be of any type and the header holds commonly required information such as timestamp, expiration, and return address. Developers can also store any arbitrary key-value properties or attributes in the header.

Since a Spring Integration Message is a generic wrapper for any Object, there is no limit to the number of potential sources for such messages. In fact, a Source implementation can act as an adapter that converts Objects from any other system into Spring Integration Messages. To facilitate the conversion of Objects to Messages, Spring Integration also defines a strategy interface for creating Messages called MessageCreator. While it is relatively easy to implement Source directly, an adapter is also available for invoking arbitrary methods on plain Objects. Also, several Source implementations are already available within the Spring Integration Adapters module. For a detailed discussion of the various adapters, see .

Just as a Source enables Message reception, a Target handles the responsibility of sending Messages. As with a Source, a Target can act as an adapter that converts Messages into the Objects expected by some other system. Spring Integration provides a strategy interface for mapping Messages to Objects called MessaegMapper. The Target interface may be implemented directly, but an adapter is also available for invoking arbitrary methods on plain Objects (delegating to the Message-mapping strategy in the process). As with Sources, several Target implementations are already available within the Spring Integration Adapters module as discussed in .

As described above, the Source and Target components support conversion between Objects and Messages so that application code and/or external systems can be connected to a Spring Integration application rather easily. However, both Source and Target are unidirectional while the application code or external system to be invoked may provide a return value. The Message Handler interface supports these request-reply scenarios. As with the Source and Target, Spring Integration also provides an adapter that itself implements the Message Handler interface while supporting the invocation of arbitrary methods on plain Objects. The adapter relies upon the message-creating and message-mapping strategies to handle the bidirectional Object/Message conversion. For more information about the Message Handler, see .

A Message Channel represents the "pipe" of a pipes-and-filters architecture. Producers send Messages to a channel, and consumers receive Messages from a channel. By providing both send and receive operations, a Message Channel basically combines the roles of Source and Target. Spring Integration provides a number of different channel implementations: QueueChannel, PriorityChannel, RendezvousChannel, DirectChannel, and ThreadLocalChannel. These are described in detail in .

Thus far, the component diagrams show Consumers, Producers, and Requesters invoking the Source, Target, and Message Handlers respectively. However, one of the primary goals of Spring Integration is to simplify the development of enterprise integration solutions through inversion of control. This means that you should not have to implement such Producers, Consumers, and Requesters directly. Instead, you should be able to focus on your domain logic with an implementation based on plain Objects. Then, by providing declarative configuration, you can "connect" your application code to the messaging infrastructure provided by Spring Integration. The components responsible for these connections are Message Endpoints. A Message Endpoint represents the "filter" of a pipes-and-filters architecture. As mentioned above, the endpoint's primary role is to connect application code to the messaging framework and to do so in a non-invasive manner. In other words, the application code should have no awareness of the Message objects or the Message Channels. This is similar to the role of a Controller in the MVC paradigm. Just as a Controller handles HTTP requests, the Message Endpoint handles Messages. Just as Controllers are mapped to URL patterns, Message Endpoints are mapped to Message Channels. The goal is the same in both cases: isolate application code from the infrastructure. Spring Integration provides three types of endpoints - one for each of the component types described above: Source Endpoint, Target Endpoint, and Handler Endpoint.

A Source Endpoint connects any Source implementation to a Message Channel. The invocation of the Source's receive operation is controlled by scheduling information provided within the Source Endpoint's configuration. Any time the receive operation returns a non-null Message, it is sent to the channel.

A Target Endpoint connects a Message Channel to any Target implementation. The invocation of the Message Channel's receive operation is controlled by scheduling information provided within the Target Endpoint's configuration. Any time a non-null Message is received from the channel, it is sent to the Target.

Since Message Handler's are capable of returning reply Messages, the Handler Endpoint has some additional responsibilities. The general behavior is the same as the Target Endpoint, but the Handler Endpoint must make a distinction between "input-channel" and "output-channel". Whenever the Message Handler does return a reply Message, that Message is sent to the output channel. If no output channel has been configured, then the reply will be sent to the channel specified as the Message header's "return address" if available.

A Message Router is a particular type of MessageHandler that is capable of receiving a Message and then deciding what channel or channels should receive the Message next. Typically the decision is based upon the Message's content and/or metadata. A Message Router is often used as a dynamic alternative to configuring the input and output channels for an endpoint.

The Message Bus acts as a registry for Message Channels and Message Endpoints. It also encapsulates the complexity of message retrieval and dispatching. Essentially, the Message Bus forms a logical extension of the Spring application context into the messaging domain. For example, it will automatically detect Message Channel and Message Endpoint components from within the application context. It handles the scheduling of pollers, the creation of thread pools, and the lifecycle management of all messaging components that can be initialized, started, and stopped. The Message Bus is the primary example of inversion of control within Spring Integration.