Pass the Salesforce Integration Architecture Designer Integration-Architect Questions and answers with ValidTests

The recommendations that the Salesforce integration architect should make to fulfill the requirements around caching, queuing, error handling, alerts, retries, event handling, etc. arebased on the best practices and capabilities of the middleware tools. The recommendation B is valid because middleware tools can provide true message queueing for integration scenarios that require asynchronous processing, guaranteed delivery, load balancing, etc. The recommendation D is valid because middleware tools can handle various events that may occur during the integration process, such as logging errors, sending notifications, triggering recovery actions, etc. The recommendation E is valid becausemiddleware tools can support the publish/subscribe pattern for event-driven integration, where the middleware can route messages from publishers to subscribers based on topics or filters. The recommendation A is not valid because transforming a fire-and-forget mechanism to request-reply does not improve performance, but rather increases the latency and complexity of the integration. The recommendation C is not valid because message transformation and protocol translation are tasks that are better suited for middleware tools than Salesforce, as they can handle different formats and protocols more efficiently and flexibly23

1: Integration Architecture Designer Resource Guide 2: Integration Patterns and Practices 3: Salesforce Integration Architecture Designer Certification Exam Guide

Create a special user solely for the integration purposes. This is the first thing that the Architect should do when building a solution that allows a service to access Salesforcethrough the API. Creating a special user for the integration purposes can help to ensure security, accountability, and traceability of the API calls. The special user should have a unique username, password, security token, and profile that grants only thenecessary permissions and access for the integration. The special user should also be assigned to an API-only user license type that prevents logging in to the Salesforce UI. References: Certification - Integration Architect - Trailhead, [User Licenses], [API User Permission]

The Fire and Forget pattern is suitable for this business use case because it allows sending a message from Salesforceto the remote system without waiting for a response or acknowledgement. This reduces the latency and complexity of the integration and enables asynchronous processing of the orders in the remote system. The platform event defined in Salesforce can be used to publish the order details to the remote system, which can subscribe to the event and create the order accordingly1

The Partner WSDL is the best option for creating an integration that might be used for more than one Salesforce organization and different metadata.The Partner WSDL is loosely typed and can reflect against any configuration of Salesforce. It is static and does not change if modifications are made to an organization’s Salesforce configuration. Therefore, it is more flexible and adaptable than the Enterprise WSDL, which is strongly typed and bound to a specific configuration of Salesforce1

References: Differences between Salesforce provided WSDL files

The typical and worst-case historical response times, and how many concurrent service calls are being placed are two considerations that an integration architect should makewhen building a scalable integration solution for synchronous callouts to “available to promise” services. These two factors can affect the performance, reliability, and availability of the integration solution, as well as the user experience of the customer checkout process. Theintegration architect should design the solution to handle high volumes of service calls, optimize the response times, handle errors and timeouts, and avoid hitting governor limits or service quotas. References: Certification - Integration Architect - Trailhead, [Callout Limits and Limitations], [Integration Patterns and Practices]

The end-to-end response time from the user’s perspective is the time elapsed between the user’s request and the user’s receipt of the final response. In the diagram, this corresponds to the sum of A and H, which are the durations of the synchronous calls from the user interface to the cloud-based application and back. The other durations (B to G) are either internal tothe cloud-based application or asynchronous calls that do not affect the user’s perception of response time. Therefore, the correct answer is D, because it represents the sum of A and H.

The HttpCalloutMock interface allows testing HTTP callouts by returning a predefined response in a test context. By implementing HttpCalloutMockto return responses per RAML specification, the Apex REST API Clients unit tests can confirm that the API requests and responses match the expected format and values. Calling the Apex REST API Clients in a test context to get the mock response is also necessary to verify the adherence to the RAML specs. Calling the HttpCalloutMock implementation from the Apex REST API Clients or requiring the Apex REST API Clients to implement the HttpCalloutMock are not valid options because the HttpCalloutMock interface is implemented by a separate class that is passed as a parameter to the Test.setMock method2

References: 1: Idempotent REST APIs 2: Testing HTTP Callouts by Implementing the HttpCalloutMock Interface

 Invoking middleware service to retrieve valid shipping methods is a solution that can allow the sales representatives to select between valid services for the customer’s country. The middleware service can act as a single point of entry for all shipping services and provide routing and transformation capabilities. Using middleware to abstract the call to the specific shipping services is a solution that can allow the sales representatives to request shipping estimates from the selected service. The middleware can hide thecomplexity and heterogeneity of the shipping services and provide mediation and orchestration capabilities. Using Platform Events to construct and publish shipper-specific events is not a solution, as Platform Events are used for event-driven integration,not for web-service integration. Storing shipping services in a picklist that is dependent on a country picklist is not a solution, as it does not address how to request shipping estimates from the shipping services. Reference: Salesforce Integration Architecture Designer Resource Guide, page 16-17

The message durability considerations that an Integration Architect should make when designing a solution to meet these business requirements are:

When subscribing to Salesforce Event bus, ReplayID is used with a value of -2 to be able to see old and new events.

High-volume event messages are stored for 72 hours (three days).

A Salesforce Event bus is a messaging service that allows you to publish and subscribe to events inSalesforce or from external sources. ReplayID is a property of an event message that indicates its position in the event stream. By using ReplayID with a value of -2, you can replay all events stored in the retention window, which can help you see the stages of order processing in Salesforce. High-volume event messages are a type of event message that can handle large volumes of data and are optimized for performance and scalability. High-volume event messages are stored for 72 hours (three days) in the event bus, which means they can be replayed within that time frame. The other options are not correct. Using ReplayID with a value of -1 means you can only see new events that are published after your subscription. High-volume event messages are not stored for 24 hours (one day), but for 72 hours (three days).

This is because Bulk API is a RESTful API that is optimized for loading or deleting large sets of data asynchronously. By processing jobs in parallel mode, you can take advantage of the multiple resources available on the Salesforce platform and speed up the data load. The other options are not suitable for this scenario because:

B, Contact Salesforce support to schedule performance load, is not a valid option as Salesforce does not offer such a service. You are responsible for optimizing your own data load performance.

C, Use Bulk API to process jobs in serial mode, is a slower option than parallel mode as it processes one batch at a time. This is only recommended when you have dependencies betweenrecords or when you want to preserve the order of records.

D, Use Bulk API to process jobs in high performance mode, is not a valid option as there is no such mode in Bulk API. You can only choose between parallel and serial modes.

References:

Introduction to Bulk API 2.0 and BulkAPI

Use Bulk API 2.0 Unit