Azure Cosmos DB AI Developer Associate (DP-420) - Full Practice Exam 1

1. A browser application uses an allowed CORS origin and assumes this authorizes all Cosmos DB requests. What should the engineer explain?
2. A business permits stale reads only within a configured number of versions or time interval. Which consistency model directly expresses bounded staleness?
3. A candidate partition key has only three possible values for a rapidly growing write-heavy dataset. What concern should the design review prioritize?
4. A catalog contains exact product terms and semantically related descriptions. Which retrieval design should be evaluated?
5. A change-feed consumer writes to an external system. What property should its processing logic have to handle possible repeated delivery safely?
6. A change-feed processor scales across several worker instances and must coordinate progress and lease ownership. Which supporting resource is part of this design?
7. A chatbot must answer from current Cosmos DB knowledge rather than rely only on a model's training. Which pattern should be used?
8. A client builds a query using a customer-supplied string. What is the best approach to avoid mixing untrusted values with query syntax?
9. A client receives a continuation token while listing query results. How should it retrieve the next page reliably?
10. A client receives HTTP 429 responses during peak load. What should its retry design do?
11. A client will perform a point read of a known item in a partitioned Azure Cosmos DB container. Which TWO values identify the target? Select TWO.
12. A Cosmos DB private endpoint has been approved. Clients still resolve the account endpoint publicly and public access is disabled. What should be investigated?
13. A customer has an unbounded history of events that is frequently appended. Why might referencing separate event items be preferable to embedding the complete history in one customer document?
14. A developer attempts one transactional batch for items with unrelated logical partition keys. What should the design review identify?
15. A Fabric or Spark workflow must read and write supported Cosmos DB data using distributed processing. Which integration should be evaluated?
16. A globally active application writes from multiple supported regions. What additional design issue must be considered?
17. A high-throughput service creates a new Cosmos SDK client for every request and experiences connection overhead. Which change should it evaluate?
18. A multitenant application uses one constant partition-key value for all tenants. Write throttling occurs despite spare aggregate capacity. What should be reconsidered?
19. A platform operates many Cosmos DB accounts and wants to evaluate centralized usage visibility and supported throughput-pooling features. Which capability should it review?
20. A query needs ordering on multiple properties. What index capability should be evaluated when required by the query pattern?
21. A shopping session must read its own recent writes through a correctly configured SDK session context. Which consistency level is designed for this requirement without demanding global strong consistency?
22. A support application needs read-only access to database items. Which authorization principle should guide its role assignment?
23. A tool masks selected sensitive fields in supported query results. Why must the application still enforce data permissions?
24. A user can manage an account through Azure Resource Manager but cannot read its items using Entra token authentication. Which permission boundary should be checked?
25. A vector query includes tenantId as a shard filter. Is that alone sufficient proof that the caller is authorized for the tenant?
26. A workload can tolerate temporary stale reads and prioritizes availability and lower read overhead over immediate latest-write visibility. Which level should be evaluated?
27. A workload rarely queries large metadata fields, but every write indexes them. What optimization should be evaluated?
28. An agent stores durable conversation state for many customers. Which key design and access control are essential?
29. An agent's semantic memory returns an old preference that the user has replaced. What should the memory design support?
30. An application changes two fields in a large item and wants to avoid sending the entire replacement document. Which operation should it evaluate?
31. An application stores document embeddings and queries semantic similarity. What must the vector policy and index match?
32. An application uses hierarchical partition keys beginning with tenantId. Queries often filter by tenantId before a more specific child key is known. What benefit should be evaluated?
33. An Azure-hosted API must read selected Cosmos DB data without storing an account key. What should be configured?
34. An ingestion job needs high throughput for many independent writes and does not require all items to commit atomically. Which approach should be considered?
35. An operator overwrites important items. The organization needs a supported point-in-time recovery within its configured retention window. Which backup capability should be evaluated?
36. An order and inventory reservation are stored as items in the same logical partition. They must succeed or fail together. Which operation should be evaluated?
37. An order and its small immutable line-item summary are usually read together. Which modeling option should be evaluated first?
38. An update fails because its If-Match ETag no longer matches. What is the appropriate application response?
39. Analysts need to explore eligible Cosmos DB data in Fabric without sending every analytical scan to the operational application path. Which integration should be evaluated?
40. Document vectors use one embedding model, but query vectors are produced by an incompatible model. What should be corrected?
41. Most requests read one tenant's recent records, while large tenants need distribution within their data. Which design should be evaluated?
42. Most throttling comes from one logical partition while overall provisioned throughput is not fully used. What is the most relevant architectural issue?
43. Most vector searches are restricted to one tenant. The team wants a supported DiskANN index design that narrows search to relevant subsets. Which capability should it evaluate?
44. One application query consumes far more RUs than expected. Which evidence is most useful before changing the schema?
45. Operations needs alerts for sustained database throttling and visibility into request latency. What should be configured?
46. Product retrieval must support text relevance over supported text paths. What should be configured for efficient full-text operations?
47. Session items should expire after a configured interval since their last modification. Which Cosmos DB capability fits?
48. Two items share the same id but have different partition-key values in one Azure Cosmos DB container. Which TWO statements are correct? Select TWO.
49. Two workers read the same item and both attempt updates. The application must reject an update based on an outdated version. Which mechanism should it use?
50. Users in different geographies read a multi-region account. Which client configuration should be evaluated to reduce avoidable cross-region latency?

Comments — 0