Domain Weighting & Distribution
Set 4 challenges your understanding of data consistency models, vector distance operators, and zero-trust container networking: Domain 1 (25%), Domain 2 (25%), Domain 3 (27%), and Domain 4 (23%).
Domain 1: 12 Qs Domain 2: 12 Qs Domain 3: 13 Qs Domain 4: 11 Qs
Architectural Trade-offs Tested in Set 4
- Cosmos DB Consistency: Strong vs. Session vs. Eventual: Strong consistency guarantees linearizable reads across all regions at the expense of doubling write latency and restricting write regions to a single master; Session consistency remains the default balance of monotonic reads within client sessions and sub-10ms latency.
- Event Hubs Capture to Blob Storage vs. Stream Analytics Ingestion: Event Hubs Capture provides zero-code, zero-maintenance batch archiving of raw telemetry directly into Avro or Parquet blobs, whereas Stream Analytics is required only when windowed aggregations or real-time filtering are needed before persistence.
- Private Endpoints vs. Service Endpoints for Key Vault: Service endpoints route traffic over the Azure backbone while keeping public DNS IP resolution; Private Endpoints allocate a private RFC 1918 IP address from your Virtual Network, completely blocking public internet reachability.
Exam Traps & Common Distractor Pitfalls
- Common Trap: Using Session consistency when external microservices require immediate visibility of newly written vectors.
Engineering Reality: Session consistency only guarantees read-your-own-writes for the specific client session token; cross-client real-time synchronization requires Bounded Staleness or Strong consistency. - Common Trap: Allowing ACA environment ingress from any IP when an Application Gateway WAF is provisioned in front.
Engineering Reality: If an Application Gateway or Azure Front Door is deployed, the Container App Environment ingress must be configured to internal-only, or restricted via IP security restrictions to the gateway's CIDR.
★ Priority Questions (Verified Real Exam Questions):
Questions marked with Priority · Real Exam are verified questions that have previously appeared on live Microsoft AI-200 certification exams. If your preparation time is tight, use the Priority · Real Exam filter below to practice these first.
Study Strategy: Understand the mathematical differences between Cosine Similarity, Dot Product, and Euclidean (L2) distance in pgvector and Cosmos DB vector indexes. Matching the metric to the embedding model's normalization is critical.