Free Salesforce Integration-Architect Practice Exam with Questions & Answers | Set: 3

In a contract-first integration strategy using RAML (RESTful API Modeling Language), the specification defines the exact structure of requests and responses. Because Salesforce unit tests cannot perform actual network callouts, the platform requires d1evelopers to use the HttpCalloutMock interface to simulate responses.

To ensure that the integration code strictly adheres to the established RAML contract, the integration architect must mandate that the HttpCalloutMock implementation returns responses that mirror the RAML specification. This means the mock must include all required fields, correct data types, and the expected HTTP status codes (e.g., 200 OK, 201 Created) as defined in the contract. By doing this, the unit tests verify that the Apex client code can successfully parse and process the specific JSON or XML payloads defined in the RAML spec.

Option A and B are technically imprecise. The Apex client does not "implement" the mock; rather, the test class provides a separate mock implementation to the runtime via Test.setMock(). The value of the integration architecture lies in the content of that mock. If the mock is designed to return contract-compliant data, then any change to the RAML that breaks the Apex code's ability to process it will be caught immediately during the testing phase. This "Mock-as-a-Contract" approach provides a safety net, ensuring that Salesforce remains compatible with external services even as those services evolve, provided the RAML is kept up to date.

When dealing with extremely large files, such as the 2.5 GB design files mentioned, an architect must consider the platform's file size limits and storage costs. Salesforce Files have a maximum upload size of 2 GB through most interfaces, making Option B technically unfeasible for a 2.5 GB file. Furthermore, storing numerous large files natively in Salesforce would lead to excessive storage consumption and costs.

The most efficient and cost-effective approach is Data Virtualization or Redirection. By creating a custom object to store the file location URL (Option C), the actual file remains in the performant on-premise file store. When the community user needs to access the design, they are redirected to the source system, which handles the massive data transfer. This fulfills the requirement to "view" the files without the overhead of moving gigabytes of data through the Salesforce infrastructure. Option A is less ideal because a 2.5 GB download over a standard Request-Reply pattern would likely lead to timeouts and a poor user experience.

In a robust event-driven architecture, "durability" is the ability of a system to retain events so that a subscriber can retrieve them even after being offline. For an architect needing to minimize data loss over "every couple of days," the selection must focus on event types that support a high retention window.

Salesforce provides two modern event types specifically designed for high-scale, durable messaging:

Change Data Capture (CDC): This automatically broadcasts changes to Salesforce records (Create, Update, Delete, Undelete). CDC events are highly durable and are retained in the event bus for 72 hours (3 days).

High-Volume Platform Events: These are custom events defined by the architect. Like CDC, High-Volume Platform Events also provide a 72-hour retention window.

Earlier versions of the Streaming API, such as PushTopic Events and Generic Events (Option C and parts of A), typically offered a 24-hour retention window, which would not satisfy the requirement of a client re-connecting "every couple of days" (potentially 48-72 hours later).

By utilizing CDC and High-Volume Platform Events, the architect can leverage the Replay ID. When the client re-connects, it can send the Replay ID of the last event it successfully processed. Salesforce then "replays" all events that occurred during the client's downtime from that specific point, up to the 72-hour limit. This mechanism ensures zero data loss for planned or unplanned outages lasting up to three days, making it the most resilient choice for the specified requirements.

In the context of data synchronization, reliability during target system outages is a critical requirement. To guarantee that notifications of record changes are stored and available for up to 3 days (72 hours), the architect should recommend Change Data Capture (CDC).

CDC is built on a high-volume event bus architecture. When a change occurs in Salesforce, a change event is published and automatically persisted in the event bus for 72 hours. This 3-day retention window is significantly longer than older technologies like PushTopics, which only retain events for 24 hours.

If the HR system or the integration middleware goes offline, no data is lost. Once the system is restored, the subscriber can retrieve all stored event messages from the bus using a Replay ID. This replayability feature allows the external system to resume processing from exactly where it left off, ensuring that every record update is eventually replicated without the need for manual data exports or custom error-handling logic on the Salesforce side.

In this scenario, the architecture involves a Remote-Call-In pattern or Batch Synchronization, where an external system (the middleware or ETL tool) initiates communication with Salesforce to update records. For any external system to securely access Salesforce APIs and perform these updates, it must be authenticated and authorized.

The Connected App is the foundational framework that allows an external application to integrate with Salesforce using APIs and standard protocols, such as OAuth 2.0 and SAML. By configuring a Connected App, the architect can define which permissions (Scopes) the middleware has, such as the ability to access data via the REST or Bulk API. This is the correct choice because the middleware needs to "log in" to Salesforce to push the phone classification data back into the Account or Contact records.

Option B, an API Gateway, is typically used to manage and secure requests going out of an organization to external services, or to provide a facade for on-premise APIs; it does not handle the inbound authentication into Salesforce itself. Option C, Remote Site Settings, is a configuration used solely to permit Salesforce to make outbound calls to a specific external URL (for example, if Salesforce were calling the phone classification service directly via Apex).

Given that the business is flexible with timing (allowing for nightly or 12-hour syncs) and handles 100,000 calls, a Batch Synchronization pattern via an ETL tool is highly efficient. The ETL tool will authenticate against the Connected App using a secure OAuth flow (such as the JWT Bearer Flow for server-to-server integration), retrieve the new phone numbers, call the external classification API, and then bulk-update the Salesforce records. This setup ensures a secure, scalable, and manageable integration that respects Salesforce's security architecture while meeting the high-volume data requirements of the call center.

For HR synchronization requiring high reliability and a multi-day data retention window, Change Data Capture (CDC) is the optimal architectural choice.12

CDC is an event-driven integration framework that automatically publ3ishes change events for Sale4sforce records (Create, Update, Delete, and Undelete). A critical feature of CDC for failure recovery is its 72-hour (3-day) retention window. If the external HR system or the replication middleware experience downtime, the change events remain stored in the Salesforce event bus for up to three days. Once the HR system is back online, the subscriber can use the Replay ID of the last successfully processed event to retrieve all missed notifications from the bus, ensuring zero data loss during the outage.

In contrast, Outbound Messaging (Option B) only provides a 24-hour retry window. If the target system is unavailable for more than one day, the notifications are dropped, making it unsuitable for a 3-day requirement. Callouts (Option C) are typically synchronous and do not have native, platform-managed storage or retry mechanisms for failed requests; once a callout fails, the data is lost unless a custom retry architecture is built. CDC’s native, durable event bus provides the necessary scalability and resilience for critical HR data synchronization without requiring complex custom code for error handling and data recovery.

To satisfy the requirement of encrypting standard fields while maintaining their functionality within record-triggered flows, Shield Platform Encryption is the recommended architectural solution.1

Shield Platform Encryption is a modern security layer that allows for encryption at rest while 2preserving critical platform features. Unlike Classic Encryption (Option B)—which is limited to a specific "Encrypted Text" custom field type and often breaks platform features like search and automation—Shield is designed to work with standard fields such as Name, Email, and Phone.

Key architectural considerations for Shield include:

Compatibility with Automation: Shield fields can be used in Flows, Apex triggers, and validation rules. This allows NTO to implement the required record-triggered business logic without needing to decrypt the data manually in code.

Search and Filtering: By using Deterministic Encryption, Shield allows users to filter and search for records based on encrypted fields, which is often a requirement for "widely-used" standard fields.

Compliance and Governance: Shield provides advanced key management (Bring Your Own Key - BYOK) and auditing, ensuring that NTO meets corporate security guidelines while data is being processed by the platform.

Data Masking (Option C) is primarily used for sandboxes to obfuscate PII during testing and is not a production encryption-at-rest solution. By recommending Shield, the architect provides a transparent security model that protects sensitive data without sacrificing the declarative power of Flow Builder.

When dealing with high-volume data (1 million transactions per day) that does not need to be stored natively in Salesforce, the architect should recommend Data Virtualization via Salesforce Connect.

Salesforce Connect allows the company to display external data as External Objects. This approach provides several architectural advantages for a banking community:

No Data Storage: Transactions remain in the core banking system, avoiding the massive storage costs and complex synchronization logic required to house millions of records natively in Salesforce.

Real-Time Visibility: Because External Objects are queried on-demand via the OData protocol or a custom Apex adapter, customers see the most up-to-date transaction history every time they refresh the page.

While an Iframe (Option B) is technically possible, it is often discouraged due to security concerns (such as clickjacking) and a poor user experience, as the Iframe does not natively integrate with Salesforce UI components or reporting. Salesforce Connect provides a "seamless" look and feel, allowing External Objects to be used in related lists and Lightning components just like standard Salesforce records, while keeping the heavy data burden on the performant core banking system.

In a complex landscape where multiple systems contain overlapping customer data, each with its own primary key, the core architectural challenge is Identity Management. To ensure that an update in Salesforce (the System of Record) correctly updates "Customer A" in the ERP and "Customer A" in the Data Lake, a Master Data Management (MDM) strategy is required.

An MDM solution creates a Cross-Reference (X-Ref) Table or a "Golden Record" that maps the unique identifiers from all systems. In the Salesforce record, the architect should implement External ID fields for each corresponding system (e.g., ERP_ID__c, Ticket_System_ID__c).

Why this is the superior recommendation:

Bidirectional Integrity: When the middleware receives an update from the ERP, it uses the ERP_ID__c to perform an "upsert" in Salesforce, ensuring no duplicates are created.

Traceability: It allows for easy auditing of data lineage across the enterprise.

Decoupling: Salesforce doesn't need to know the internal logic of the ERP; it simply holds the reference key.

Option B (CDC) is a delivery mechanism, not an identity management strategy; it tells you that something changed, but not which record in the ERP it corresponds to without the ID mapping. Option C (Local caching in middleware) is an "anti-pattern" because it makes the middleware stateful; if the middleware cache is lost or out of sync, the entire integration breaks. By designing an MDM-based mapping solution directly within the data model, the architect ensures a robust, scalable, and transparent identity framework for the entire enterprise.

In a multi-system landscape, maintaining data synchronization requires a robust Identity Mapping strategy. The standard Salesforce architectural recommendation is to use External ID fields to store the unique identifiers from each secondary system.

By storing the ERP ID, Ticketing ID, and Data Lake ID as External IDs in Salesforce, the middleware can perform upsert operations without needing to first query Salesforce for its internal ID. This reduces the number of API calls and simplifies the integration logic. Conversely, when Salesforce pushes a change to the ERP, it sends the stored ERP ID, allowing the ERP to instantly identify the correct target record.

Option B (Caching at the middleware) is a high-maintenance "anti-pattern" that introduces a new point of failure if the cache goes out of sync with the actual systems. Option C (Change Data Capture) is a mechanism for notifying systems of changes, but it does not solve the underlying identity mapping problem. Using External IDs creates a stable, searchable, and performant cross-reference that is the backbone of any successful "hub-and-spoke" integration architecture.