Microsoft Certified: Azure Cosmos DB Developer Specialty Practice Exam 2025: Latest Questions

You are designing a new Azure Cosmos DB for NoSQL container to store customer orders. Most queries retrieve a single order by orderId and include line items. You need to minimize request units (RUs) and avoid cross-partition queries when retrieving a single order. Which data modeling approach should you use?

A web application uses Azure Cosmos DB for NoSQL. The application frequently retrieves a document when both the id value and the partition key value are known. Which operation provides the lowest latency and RU consumption for this access pattern?

You are troubleshooting slow queries in an Azure Cosmos DB for NoSQL container. You want to identify whether queries are using the index or performing full scans. What should you review first?

Your team needs to automatically delete IoT telemetry documents after 7 days to reduce storage and keep queries fast. What is the recommended Cosmos DB feature to implement this requirement?

A container uses /customerId as the partition key. During peak periods, RU consumption is high due to frequent queries that filter by orderStatus and orderDate across all customers. You need to reduce RU cost for these queries without changing the partition key. What should you do?

You are designing a multi-tenant SaaS solution in Azure Cosmos DB for NoSQL. Each tenant has highly variable load, and you must ensure no single tenant can exhaust throughput for others. The solution must still support efficient queries within a tenant. Which partition key strategy is most appropriate?

A global application uses Azure Cosmos DB for NoSQL with multiple write regions enabled. Users occasionally report seeing conflicts where the same document is updated in two regions at nearly the same time. You must ensure deterministic conflict resolution where the highest application-defined version wins. What should you implement?

Your application ingests events into a Cosmos DB for NoSQL container. A downstream service must process each new or updated item in near real time and maintain state between runs. Which integration approach is recommended?

A container experiences intermittent 429 (Too Many Requests) responses during traffic spikes. You must reduce end-user impact while keeping costs predictable and avoiding over-provisioning manual throughput. What should you implement?

You need to design a container for user activity logs. The primary query pattern is: retrieve the most recent 100 activities for a specific userId, ordered by activityTime descending. The system must scale to high write volumes without creating hot partitions. Which design is most appropriate?

You are designing an Azure Cosmos DB for NoSQL container to store IoT device telemetry. Each device sends one event per second. Your most common query is: "get the last 5 minutes of events for a specific device, ordered by timestamp". Which partition key choice best supports this access pattern while avoiding cross-partition queries?

A team wants to prevent their client application from requiring Azure AD to access Azure Cosmos DB. They want the app to use a short-lived credential that only grants read access to a single container. Which feature should you use?

You need to ensure that a web API uses the closest Azure Cosmos DB region for reads while still writing to the configured write region. Which configuration should you enable on the Cosmos DB account?

A container uses autoscale throughput. You observe frequent 429 (Request rate too large) responses during predictable daily peaks, even though the average load is low. Which action is most likely to reduce 429s while keeping costs aligned with usage?

You must store customer profiles and their orders in Azure Cosmos DB for NoSQL. The API frequently retrieves a customer and the customer’s last 10 orders together. Updates to orders are independent and can be high volume. Which data modeling approach is recommended?

A container uses a composite index on (deviceId ASC, timestamp DESC). A developer reports that a query with ORDER BY timestamp DESC is still failing unless they also include ORDER BY deviceId. Why?

Your application reads the same reference documents (for example, country codes) repeatedly and they rarely change. You want to reduce RU consumption and latency without changing the data model. What should you implement in the application?

You need to integrate Azure Functions to process changes from an Azure Cosmos DB for NoSQL container exactly once per change in each function instance, even across restarts. Which approach should you use?

A container is partitioned by /tenantId. Your workload has a small number of tenants generating extremely high throughput, causing throttling for those tenants while others are idle. You cannot change the partition key due to existing clients. Which mitigation is most appropriate?

You run a query that filters by partition key and an additional property. The query is still consuming high RU because it scans many items within the partition. You need to reduce RU for this point lookup-like pattern. Which change is most effective?

You use the Azure Cosmos DB .NET SDK to query items from a container. The query should only return a lightweight projection (id and status). However, RU consumption is higher than expected, and the SDK appears to be downloading full documents. Which SDK configuration should you use to reduce network payload and RU for the query results?

You are designing an analytical reporting solution. Data is stored in Azure Cosmos DB for NoSQL and must be queried with low impact to transactional workloads. You want to run near-real-time analytics without building an ETL pipeline. What should you enable?

A container uses hierarchical partition keys. You frequently run queries scoped to a single tenant and a single device within that tenant. Which partition key definition best supports this access pattern while allowing you to efficiently target both levels?

An application must enforce a unique constraint: within each tenant, the value of userName must be unique. Tenants are isolated by tenantId and all user documents include tenantId and userName. What is the best way to enforce this constraint in Azure Cosmos DB for NoSQL?

You run a high-throughput ingestion service using the Azure Cosmos DB for NoSQL SDK. You observe intermittent 429 (Request Rate Too Large) responses even though overall RU/s seems sufficient. Telemetry shows that most 429s occur when writing documents for a single customer during traffic spikes. The container is partitioned by /customerId. What change best addresses the root cause?