The Integration Engine processes the message in a step by step (pipeline) process. The following are the most important steps executed by the central pipeline:
PLSRV_RECEIVER_DETERMINATION and PLSRV_INTERFACE_DETERMINATION
The Sender CC is just the starting point. Based on the CC and the message, the group of objects (interface) involved in the processing of the message and the destination where the message ends up must be determined in XI.
The receiver determination and interface determination steps find out the details of the receiver system and the interface for a message. The receiver can be a constant one or calculated dynamically based on the payload.
If the message is being sent to multiple receivers, it will be branched to its multiple destinations at this step. The term split should not be mistook as partitioning of the message contents.
The message may require changes to its structure or values to meet the design needs. These structural/value changes are done by mapping objects. The mapping can be done in ABAP or Java.
Typically, mappings are designed to be carried out in Java as it is faster and design is relatively easy.
If Java mapping is used, the message is sent from IE to AE via RFC Program ID AI_RUNTIME_JCOSERVER (SM59).
Not all messages require changes in their structure or values. If there is no change, mapping will not happen.
PLSRV_OUTBOUND_BINDING and PLSRV_CALL_ADAPTER
These are the stages where message is handed over to its receiver adapter (or to BPE for further processing) from where it will be posted to its receiver system(s).
A successful execution of these steps concludes the pipeline processing in Integration Engine.