Integration pattern selection for health data exchange
Key Takeaway
Choose a pattern by combining direction, integration type, and timing. A regular, large load into Salesforce that must not slow users points to Batch Data Synchronization with an ETL tool and the Bulk API.
Integration patterns are proven, reusable ways to connect Salesforce with other systems, such as an external system that holds clinical, claims, or order records. Each pattern describes how systems and their components interact in an established solution, so you can solve a common integration problem without designing a new solution. A pattern gives a recommended approach and names specific technologies, such as connectors or services, to inform your choice, but it is not a step-by-step implementation guide. You choose a pattern by matching the customer's needs: which system starts the interaction, whether a reply is needed right away, how much data moves, and whether the data must be copied into Salesforce at all.
Integration type, timing, and direction
You describe an integration by its type, its timing, and its direction. The integration type says what the integration connects or shares.
- Process integration connects the functional flow of work across two or more applications. Process integration tends to be more complex, especially for transactions and rollback.
- Data integration shares the information applications use. The work ranges from a simple insert or upsert to complex updates that need referential integrity.
- Virtual integration lets Salesforce work with data that stays in an external system, without copying the data into Salesforce.
Timing is either synchronous or asynchronous. A synchronous (blocking) call sends a request and waits for the response before going on. An asynchronous (nonblocking) call sends a one-way message, and the caller moves on without waiting. Results and faults come back through other one-way operations.
Direction names which system starts the interaction: Salesforce → System or System → Salesforce. Direction is not about which way the data flows. If Salesforce calls out because of an event and the external system replies with important data, the direction is still Salesforce → System.