INTEGRATION-ARCHITECTURE-DESIGNER practice question 3 of 25

Northern Trail Outfitters submits orders to the manufacturing system web-service. Recently, the system has experienced outages that keep service unavailable for several days. What solution should an architect recommend to handle errors during these types of service outages?

Full set: 106 questions

Domain Design an Integration Solution

Answer options
A: Use middleware queuing and buffering to insulate Salesforce from system outages. Correct answer
B: A Use Platform Event replayld and custom scheduled Apex process to retrieve missed events.
C: Use @future jobld and custom scheduled apex process to retry failed service calls.
D: Use Outbound Messaging to automatically retry failed service calls.
Correct Answer:
A. Use middleware queuing and buffering to insulate Salesforce from system outages.
Explanation

The most robust and scalable solution for handling prolonged outages of an external system is to use middleware. A middleware platform can act as an intermediary, accepting the order submission from Salesforce and placing it into a durable queue. This immediately frees up the Salesforce process. The middleware then manages the delivery to the manufacturing system, retrying for as long as necessary (days, if configured) until the system is available. This pattern, often called queuing or buffering, effectively decouples Salesforce from the availability of the downstream system, ensuring no data is lost during extended outages.

Why Incorrect

B. Use Platform Event replayId and custom scheduled Apex process to retrieve missed events. Platform Events have a maximum retention of 72 hours. An outage of "several days" could exceed this limit, leading to data loss.
C. Use @future jobId and custom scheduled apex process to retry failed service calls. This requires building a complex and stateful custom retry mechanism within Salesforce, which is difficult to maintain, scale, and is prone to hitting governor limits.
D. Use Outbound Messaging to automatically retry failed service calls. Outbound Messaging only retries for a maximum of 24 hours. If the outage lasts longer, the messages will be dropped, resulting in data loss.

References

1. Salesforce Integration Patterns and Practices Guide: The "Fire-and-Forget" pattern discusses the use of a reliable messaging bus (middleware) to ensure message delivery even when the remote system is unavailable. This decouples the systems and provides the necessary durability. (See the "Fire-and-Forget" section in the guide for context on asynchronous messaging and reliability).

2. Salesforce Help - Outbound Messaging: The official documentation states, "If the endpoint is unavailable, messages stay in the queue and we retry sending for up to 24 hours. After 24 hours, we drop the message from the queue." This directly confirms that this option is unsuitable for outages lasting "several days."

3. Platform Events Developer Guide - Platform Event Retention: The guide specifies, "By default, platform events are stored for 24 hours... You can get up to three days (72 hours) of event retention with the High-Volume Platform Events add-on subscription." This highlights the risk of data loss for outages exceeding this window.

4. Enterprise Integration Patterns by Gregor Hohpe and Bobby Woolf: While not a Salesforce-specific document, this is a foundational academic text in integration architecture. Chapter 3, "Messaging Systems," and Chapter 4, "Messaging Channels," describe the use of a Message Queue to create a "time-independent" integration, allowing senders and receivers to operate without being simultaneously available. This is the core principle behind the correct answer.

Scroll to Top