Browse Certification Practice Tests by Exam Family

Microsoft DP-420 Cheat Sheet: Cosmos DB Developer

May 1, 2026

Review the Microsoft Azure Cosmos DB Developer (DP-420) scope, data modeling, partition keys, consistency, indexing, SDK integration, performance, security, and maintenance traps before practicing.

On this page

DP-420 is a Cosmos DB developer exam. Use this cheat sheet to keep design decisions tied to partitioning, data model, consistency, indexing, query performance, integration, security, and maintenance.

Use this with practice. Review the Cosmos DB checkpoints, then return to the DP-420 exam page for sample questions and update tracking.

Open DP-420 practice page Compare Azure routes

Exam snapshot

FieldDetail
IssuerMicrosoft
Certification laneAzure Cosmos DB Developer Specialty
Exam codeDP-420
Main scopeCosmos DB data modeling, partitioning, query behavior, performance, security, integration, and maintenance
IT Mastery statusSample questions available

Cosmos DB map

AreaWhat to knowCommon trap
Data modelingDocument shape, embedded vs referenced data, access patterns, and denormalizationDesigning like a relational schema without query patterns
PartitioningPartition key choice, distribution, hot partitions, tenant patterns, and scaleChoosing a low-cardinality or constant partition key
Consistency and distributionConsistency levels, replication, latency, availability, and global distributionChoosing strongest consistency without considering latency and availability
Indexing and queriesIndex policy, query efficiency, RU cost, filters, and projectionsTreating every slow query as a throughput-only issue
SDK and integrationClient configuration, retries, change feed, transactions, and service integrationIgnoring retry and throttling behavior in app design
Security and maintenanceKeys, RBAC, network access, backup, monitoring, alerts, and lifecycleSecuring the app while leaving data access broad

Must-know distinctions

DistinctionHow to decide
Partition key vs idThe partition key distributes data and workload; id uniquely identifies an item within context.
Hot partition vs high RU useA hot partition concentrates load; high RU use can also come from inefficient queries.
Embedded vs referenced dataEmbed when data is read together; reference when independent update or growth matters.
Strong vs eventual consistencyStrong favors freshness; eventual favors availability and latency tradeoffs.
Change feed vs queryChange feed reacts to item changes; queries retrieve current data by criteria.

High-yield checklist

  • Start with access patterns before choosing document shape.
  • Choose partition keys for distribution and common query filters.
  • Watch for hot partitions and cross-partition query costs.
  • Use indexing intentionally instead of accepting every default.
  • Match consistency to application tolerance for staleness and latency.
  • Design retry behavior and throttling handling in the client.
  • Use monitoring to distinguish RU pressure, hot partition, and query inefficiency.
  • Secure data access with identity, keys, network controls, and least privilege.

Common traps

  • Choosing /id as a partition key without checking distribution.
  • Modeling every relationship like a relational database.
  • Increasing throughput before optimizing the query or index.
  • Ignoring consistency requirements in global apps.
  • Forgetting change feed for event-driven integration.

Practice strategy

For DP-420 misses, write the workload access pattern first. Then decide whether the problem is document model, partition key, consistency, indexing, SDK behavior, security, or monitoring.

Revised on Monday, May 25, 2026