Choosing the integration pattern for a requirement
How to pick the integration pattern from the requirement itself: who initiates, whether anyone waits, whether the data must live in Salesforce and how much of it moves; when to virtualize instead of replicate, how to give a remote system an asynchronous call-in, which event-driven pattern fits, and what replaces Salesforce to Salesforce.
Given a use case, identify the integration pattern that meets business requirements.
~8 min read · 17 practice questions ·
What the exam expects
- Pick the pattern from a requirement's timing, volume, initiator and storage constraint, rejecting patterns that break a stated constraint.
- Decide between Data Virtualization and Batch Data Synchronization when reporting, aggregation or whole-data-set processing is involved.
- Design an "asynchronous call-in" for a remote system that can't wait, and know when a direct API update is simpler.
- Combine several patterns in one solution, one per data flow.
- Choose among the event-driven patterns (publish/subscribe, fanout, queuing, passed messages, streaming) and recognise when event-driven doesn't fit.
- Recommend the right replacement for Salesforce to Salesforce.
Key terms: initiator (the system that starts the interaction, which fixes the pattern family); Data Virtualization (Salesforce reads external data in real time without persisting it); asynchronous call-in (a remote system publishes a platform event instead of waiting on Salesforce processing); fanout (all subscribers read the same message from one queue); queuing (each subscriber has its own queue that holds messages until retrieved).
Key facts
- Detail
- Always synchronous request-reply; SOAP API has no asynchronous mode
- Detail
- Events for a variable, non-finite set of subscribers; direct SOAP/REST for simple updates to a finite set of records
- Detail
- Flow-driven platform events, consumed externally through Pub/Sub API
- Detail
- Primarily small volume, real-time; external objects can impact report performance
- Detail
- Configurable, up to 120 seconds
- Detail
- Pull-based: the client requests events
- Detail
- Unique queue per subscriber; delivery can be guaranteed and tracked
- Detail
- Single queue, same message; harder to verify receipt
- Detail
- Retired Spring '27; multi-org replacement is Data Cloud One
Sources for this section (4)
- Why is Batch Data Synchronization better than Data Virtualization for a nightly calculation over 20 million external rows?
- When should a remote system publish a platform event instead of updating records through REST?
- When is an event-driven pattern the wrong choice?
The questions are free to read. Answers with their sources and self-grading are included with the pack.
Keep reading with the Integration Architect pack
- Core concepts
- Exam traps
- Worked scenario
The pack includes:
- All 25 lessons of the Integration Architect guide, with reading progress
- Interactive decision walks and step-by-step sequence steppers
- Inline practice questions and one-click topic practice (17 questions for this lesson)
- An official source cited for every claim
- An AI tutor that answers from the guide
- Unlimited timed practice exams