Platform Integration Architect Advanced Practice Exam: Hard Questions 2025

A global manufacturer is documenting its current landscape. Salesforce orgs integrate with an on-prem ERP through a middleware layer. The ERP team reports that duplicates and occasional data loss occur only during peak order hours. The middleware logs show intermittent timeouts, and Salesforce debug logs show mixed patterns of synchronous callouts and asynchronous jobs. The architect must identify the most critical gap in the current landscape assessment to explain the symptoms before proposing a redesign. What should be investigated first?

A financial services company wants near-real-time visibility in Salesforce of payment status changes originating in a core banking system. Business stakeholders insist Salesforce must always show the latest status within 30 seconds, while the banking system cannot accept inbound calls and can only publish outbound events. The integration team also must meet audit requirements: prove that every published status change was processed exactly once, and failures must be replayable without manual data edits. What is the best integration requirement to capture from the business needs?

A retailer uses Salesforce for customer service and an external Order Management System (OMS) as the system of record for orders. Service agents need to view order line items in Salesforce and occasionally update the delivery instructions field in the OMS. During holiday season, agents may open hundreds of cases per minute, each requiring the latest order details. The OMS can handle high read volume but has strict rate limits for writes and requires strong validation. What integration requirements best translate these needs while minimizing risk?

A healthcare provider must integrate Salesforce with multiple EMR systems. Each EMR has a different patient identifier, and some systems occasionally resend the same update event with slight field differences due to late-arriving corrections. Salesforce must maintain a longitudinal patient record with a single golden Patient key, preserve a full change history for compliance, and avoid overwriting newer values with late-arriving older updates. What is the most appropriate integration requirement to prevent data integrity issues?

A SaaS company needs to expose Salesforce data and actions to a partner portal hosted outside Salesforce. Partners must search Accounts/Contacts (read) and submit service requests (create/update) with strong security boundaries: each partner can access only their own customers, and the partner system must not store Salesforce credentials. The partner portal needs low-latency reads and reliable writes with clear error handling. Which architecture best meets these requirements?

An enterprise uses an iPaaS to orchestrate integrations. A new process requires Salesforce to initiate an order submission to an external fulfillment API during an Opportunity stage change. The external API has variable latency (200 ms to 20 seconds) and occasional 503s, and the business requires that Salesforce users never experience UI delays. The submission must be retried safely without creating duplicate orders, and failures should be visible to support teams. What design is most appropriate?

A media company integrates Salesforce with a data lake. Downstream analytics requires a complete, ordered stream of Account and Contact changes, including deletes and field history-like tracking, and must support reprocessing from a point in time. The integration must not overwhelm Salesforce during spikes and must minimize custom code. Which solution is the best fit?

A team implemented a Platform Event-driven integration where an external system publishes events to Salesforce via API, and Salesforce processes them with an Apex trigger that upserts records. During load tests, some events fail with UNABLE_TO_LOCK_ROW and others exceed CPU time. The external system retries immediately on any non-2xx response, causing message storms and duplicate processing attempts. What build-time change best stabilizes processing while preserving at-least-once delivery?

An integration uses REST APIs from middleware to upsert Contacts into Salesforce. After a migration, error rates increase with intermittent 401 errors and occasional successful calls using the same client configuration. Investigation shows multiple middleware nodes share the same connected app and refresh token store. What is the most likely root cause and best fix?

A company runs multiple integrations (CDC consumers, Platform Event publishers, and REST APIs). Support reports sporadic missing updates in the downstream system, but no clear errors. The architect must improve maintainability and operability: faster root-cause analysis, end-to-end traceability across systems, and proactive detection of lag/backlogs without introducing significant overhead. What is the best approach?