Introducing the Pipeline

This section highlights the basic features to develop a new integration flow in ASB.

Integration flows are called "pipelines" in ASB.

Unlike other synchronous ESBs, the integration pipelines are message centric, and asynchronous. This means that each ASB pipeline could be asynchronously processing thousands of messages a second, and various messages could be under active processing in various nodes of the same pipeline.

Each message flows through the pipeline, and the flow itself is not sequential.

The route nodes in the pipeline declare their dependencies, and, optionally, exclusions. Once an incoming message has satisfied the dependencies listed, it is routed to the node or nodes whose dependencies are now satisfied for message flowing through the pipeline.

No assumption can be made about message sequencing, the message routing is decided by dependencies.

This allows for massive performance and deep data integrity at the message level.

Once a message has flown through all the possible nodes, as indicated by their dependencies, then the processing is complete and the message is garbage collected by the ESB engine.

This asynchronous, message oriented processing, enables the ASB to act as an API service bus, as APIs are message centric as well. The same also implies that ASB is a streaming service bus, as messages are streamed through the pipelines, being transformed on their way through the pipelines.