Microsoft · Practice Exam · Azure Cosmos DB · Specialty

Microsoft DP-420: Designing and Implementing Cloud-Native Applications Using Azure Cosmos DB Practice Exam

Cover all five DP-420 skill areas — data modeling and partitioning, data distribution, service integration, optimization, and maintaining the solution — with developer-focused scenario questions, immediate feedback in Learn mode, and full timed simulation in Exam mode.

Start 24-hour free trial →
500+
Practice questions
5
Skill areas
2
Study modes
24h
Free trial

DP-420 exam at a glance

Vendor
Microsoft (Azure)
Exam code
DP-420 (Designing and Implementing Cloud-Native Applications Using Microsoft Azure Cosmos DB)
Certification
Microsoft Certified: Azure Cosmos DB Developer Specialty
Level
Specialty (single-technology; developer)
Blueprint
Skills measured as of April 20, 2026 (current edition)
Format
Typically 40–60 questions; multiple choice, multiple response, drag-and-drop, and case studies
Duration
About 100–120 minutes
Passing score
700 of 1000 (scaled, not a raw percentage)
Delivery
Pearson VUE test center or online proctored
Recommended experience
Developing apps for Azure and working with Azure Cosmos DB for NoSQL; writing efficient SQL queries, indexing policies, and JSON; reading C# or Java; server-side JavaScript; PowerShell
Cost (USD)
$165 USD list price (specialty tier; varies by region)
Validity
Valid one year; renewed free via an online assessment on Microsoft Learn

Source: Microsoft Learn — DP-420 study guide (skills measured) and the Azure Cosmos DB Developer Specialty certification page. Verify current details with Microsoft before scheduling.

About the Microsoft Certified: Azure Cosmos DB Developer Specialty certification

DP-420 validates that you can design and implement cloud-native applications on Azure Cosmos DB for NoSQL. It is a developer exam, not an administration one: expect to reason about data models and partition keys, write efficient SQL queries and indexing policies, use the SDKs (C# or Java) for point operations, transactional batches, and change feed, write server-side JavaScript, and design for consistency, cost (request units), and global distribution. A single exam earns the Azure Cosmos DB Developer Specialty certification.

DP-420 is an active exam (current edition dated April 20, 2026, with only minor changes from the prior version). It is a Specialty credential — a focused, single-technology certification distinct from the broad associate exams. It is not an entry point: Microsoft expects real Azure development experience. If you are new to Azure data services, start with DP-900 Azure Data Fundamentals, and if you are building your Azure development foundation, AZ-204 Azure Developer Associate is the natural companion.

Every PowerKram practice question maps to one of the five DP-420 skill areas and links to the specific Microsoft Learn page it was derived from, so your weak spots become a focused reading list rather than a guess.

DP-420 skill areas and weights

Five skill areas with an unusually wide spread: data modeling dominates the exam at 35–40%, and maintaining the solution is a strong second at 25–30%, while data distribution and integration are light at 5–10% each. Microsoft published each area as a weighting range; the ranges below are reproduced as Microsoft stated them. Plan your study time in proportion.

Design and implement data models

The core of the exam: design non-relational models for Cosmos DB for NoSQL (embedding vs referencing, denormalization, TTL, schema versioning); design a partitioning strategy (partition key choice, synthetic and hierarchical keys, cross-partition cost); size and scale (serverless vs provisioned vs free tier); SDK connectivity; SQL data access (arrays, subqueries, functions); SDK operations (point vs query, transactional batch, bulk, ETags, pagination, 429 handling); and server-side JavaScript (stored procedures, triggers, UDFs).

35–40%Heaviest
Design and implement data distribution

Replication strategy (when to distribute, automatic failover policies, manual failover of write regions); choosing and evaluating consistency models against availability, RU cost, performance, and latency; and multi-region writes with custom conflict-resolution policies.

5–10%
Integrate an Azure Cosmos DB solution

Enable analytical workloads (Cosmos DB mirroring for Microsoft Fabric, analytical store, Synapse Link, change data capture); and implement solutions across services using Change Feed with Azure Functions and Event Hubs for denormalization, referential integrity, aggregation, and archiving, plus Azure AI Search integration.

5–10%
Optimize an Azure Cosmos DB solution

Optimize query performance (adjust indexes, calculate query and point-operation RU cost, integrated cache); design and implement change feeds (Functions triggers, SDK consumption, change feed estimator); and define an indexing strategy (read- vs write-heavy, index types, composite indexes, custom policies).

15–20%
Maintain an Azure Cosmos DB solution

The second-heaviest area: monitor and troubleshoot (Normalized RU consumption, server-side latency, replication, Azure Monitor alerts and resource logs, partition distribution, security auditing); backup and restore (periodic vs continuous, point-in-time restore); security (service- vs customer-managed keys, network access, RBAC control and data plane, Entra ID, CORS, Key Vault, Always Encrypted); data movement (bulk, Data Factory, Kafka, Stream Analytics, Spark connector); and DevOps (ARM templates, throughput migration, regional failover).

25–30%Second heaviest

Source: Microsoft Learn — DP-420 study guide (skills measured, as of April 20, 2026). Microsoft published weightings as ranges; data modeling is the single heaviest area at 35–40%.

Who DP-420 is for

Azure Cosmos DB Developer Specialty is aimed at developers who build applications on Cosmos DB:

  • Application developers designing data models and writing data-access code against Cosmos DB for NoSQL with the C# or Java SDKs.
  • Cloud-native and backend engineers building globally distributed, low-latency services that integrate Cosmos DB with Azure Functions, Event Hubs, and AI Search.
  • Data platform developers tuning partitioning, indexing, request units, and consistency for cost and performance.
  • DevOps-minded developers provisioning and maintaining Cosmos DB with ARM templates, PowerShell, and Azure CLI.

DP-420 is not an entry point — Microsoft expects real Azure development experience, comfort reading C# or Java, and server-side JavaScript. If you are new to Azure data services, the DP-900 Azure Data Fundamentals exam is the conceptual starting point, and AZ-204 Azure Developer Associate builds the broader Azure development foundation. For the roles this credential supports — with skills, tools, and salary ranges — see the Data Engineer career path.

What this DP-420 practice exam delivers

Learn mode

Get the correct answer, the reasoning, and a direct link to the exact Microsoft Learn page each question was derived from — immediately after each question. Built for the data-modeling and partitioning scenarios that dominate DP-420.

Exam mode

A timed run in the DP-420 shape — roughly 40 to 60 questions in about 100 to 120 minutes, including case studies — so you build pacing for the exam’s developer-scenario format.

Source-linked explanations

Every answer cites the exact Microsoft Learn documentation page it was built from — so you can verify Cosmos DB behavior around partitioning, RU cost, consistency, and the SDKs, not just memorize.

Score by skill area

Results break down by the five DP-420 areas — data models, distribution, integration, optimization, and maintenance — so practice tells you exactly which area to revisit. Expect to spend the most time on data modeling.

Sample DP-420 practice questions

Ten free questions across the five DP-420 skill areas, weighted toward data modeling as the real exam is, with full explanations and source links to the Microsoft Learn pages each is derived from. The complete bank is available with the 24-hour trial.

Question 1 · Data models (35–40%)

You are choosing a partition key for a high-volume IoT telemetry container. Queries almost always filter by device, and writes are spread evenly across millions of devices.

Which partition key characteristic should you prioritize?

  1. High cardinality that spreads requests evenly and aligns with the common query filter
  2. A single fixed value so all data lands in one partition
  3. A random GUID unrelated to any query filter
  4. A low-cardinality status field with three possible values
Show answer & explanation

Correct: A — high cardinality that spreads requests evenly and aligns with the common query filter. A good partition key distributes storage and throughput evenly and matches the dominant access pattern (here, device), avoiding hot partitions and cross-partition queries — exactly what this workload needs.

Why not the others: a single fixed value (B) and a low-cardinality status field (D) concentrate load into few partitions (hot partitions); a random GUID (C) spreads writes but forces expensive cross-partition queries because it does not match the query filter.

Source: Microsoft Learn — Partitioning in Azure Cosmos DB → Further reading: PowerKram — DP-900 Azure Data Fundamentals →
Question 2 · Data models (35–40%)

A multitenant app queries by tenant and then by user within a tenant. You want efficient, targeted queries at both levels without creating a hot partition per tenant.

Which partitioning approach should you use?

  1. A single system-wide partition key
  2. A hierarchical partition key (for example, tenantId then userId)
  3. No partition key at all
  4. Storing every tenant in its own database account
Show answer & explanation

Correct: B — a hierarchical partition key (tenantId then userId). Hierarchical (sub-)partition keys let Cosmos DB target queries efficiently at multiple levels and avoid a single large logical partition per tenant — exactly the multitenant pattern described.

Why not the others: a single system-wide key (A) creates hot partitions; no partition key (C) is not viable at scale; a database account per tenant (D) is heavy operational overhead and is not the partitioning mechanism the exam targets here.

Source: Microsoft Learn — Hierarchical partition keys →
Question 3 · Data models (35–40%)

Your SDK application intermittently receives HTTP 429 responses during traffic spikes on a provisioned-throughput container.

What does a 429 indicate, and what is the correct handling?

  1. A permission error; rotate the account keys
  2. A request-rate-too-large (throttling) error; retry after the recommended interval or increase RU/s
  3. A network timeout; switch to gateway mode permanently
  4. A schema-validation failure; change the document shape
Show answer & explanation

Correct: B — a request-rate-too-large (throttling) error; retry after the recommended interval or increase RU/s. A 429 means the request exceeded provisioned request units; the SDK surfaces a recommended retry-after, and you either let the SDK retry or provision more throughput (or use autoscale) — exactly the correct handling.

Why not the others: a 429 is not a permission error (A), a network timeout (C), or a schema failure (D) — those are unrelated conditions with different remedies.

Source: Microsoft Learn — Request units in Azure Cosmos DB → Further reading: PowerKram — AZ-204 Azure Developer Associate →
Question 4 · Data models (35–40%)

You need to run custom business logic that atomically creates and updates several items in the same logical partition in a single transaction, executed on the server.

Which Cosmos DB for NoSQL feature should you use?

  1. A user-defined function (UDF)
  2. A change feed processor
  3. A stored procedure
  4. An analytical store query
Show answer & explanation

Correct: C — a stored procedure. Stored procedures run server-side JavaScript and execute atomically within a single logical partition, making them the right tool for multi-item transactional logic on the server — exactly the requirement.

Why not the others: a UDF (A) only computes values inside queries and cannot write; a change feed processor (B) reacts to changes asynchronously, not atomically; an analytical store query (D) reads the column store for analytics and does not write transactionally.

Source: Microsoft Learn — Stored procedures, triggers, and UDFs →
Question 5 · Data distribution (5–10%)

A globally distributed app needs the strongest possible consistency guarantee, accepting higher latency and RU cost in exchange for linearizable reads.

Which Cosmos DB consistency level should you choose?

  1. Eventual
  2. Session
  3. Consistent prefix
  4. Strong
Show answer & explanation

Correct: D — Strong. Strong consistency provides linearizable reads — readers always see the most recent committed write — at the cost of higher latency and RU consumption, and it constrains write-region topology. That matches the requirement exactly.

Why not the others: Eventual (A) is the weakest and lowest-cost; Session (B) guarantees consistency only within a client session; Consistent prefix (C) guarantees ordering but not the latest value. None gives linearizable reads.

Source: Microsoft Learn — Consistency levels → Further reading: PowerKram — DP-900 Azure Data Fundamentals →
Question 6 · Integrate (5–10%)

You must react to every insert and update in a Cosmos DB container to keep a downstream search index and an aggregate summary up to date, in near real time.

Which Cosmos DB capability should you build on?

  1. A periodic backup restore
  2. A composite index
  3. The change feed, consumed via an Azure Functions trigger
  4. A manual regional failover
Show answer & explanation

Correct: C — the change feed, consumed via an Azure Functions trigger. The change feed emits a persistent, ordered record of inserts and updates; an Azure Functions trigger processes it to drive downstream denormalization, aggregation, and search indexing in near real time — exactly the integration described.

Why not the others: a backup restore (A) is a recovery operation; a composite index (B) speeds queries but does not stream changes; a manual regional failover (D) is an availability action. None reacts to item changes.

Source: Microsoft Learn — Change feed in Azure Cosmos DB → Further reading: PowerKram — AZ-204 Azure Developer Associate →
Question 7 · Optimize (15–20%)

A write-heavy container rarely queries most properties, yet every write consumes high RU because all paths are indexed. You want to cut write RU cost.

What is the most effective indexing change?

  1. Index every property with a composite index
  2. Switch the account to Strong consistency
  3. Customize the indexing policy to exclude paths that are never used in queries
  4. Enable continuous backup
Show answer & explanation

Correct: C — customize the indexing policy to exclude paths that are never used in queries. For write-heavy workloads, excluding unqueried paths from the indexing policy reduces the indexing work per write and therefore write RU cost — exactly the optimization the scenario needs.

Why not the others: indexing every property (A) increases write RU; changing consistency (B) affects reads and availability, not indexing cost; continuous backup (D) is a recovery setting with no effect on write RU.

Source: Microsoft Learn — Indexing policies →
Question 8 · Optimize (15–20%)

A workload has a spiky, unpredictable traffic pattern with long idle periods and short bursts. You want to avoid over-provisioning while still absorbing bursts.

Which throughput option best fits, all else equal?

  1. Manually provisioned fixed RU/s sized for peak
  2. Autoscale (or serverless) throughput that adapts to demand
  3. Disabling indexing entirely
  4. Strong consistency
Show answer & explanation

Correct: B — autoscale (or serverless) throughput that adapts to demand. For spiky, unpredictable traffic with idle periods, autoscale scales RU/s up for bursts and down when idle (and serverless bills per request), avoiding paying for peak capacity around the clock — exactly the cost-optimization goal.

Why not the others: fixed RU/s sized for peak (A) wastes capacity during idle periods; disabling indexing (C) breaks queries; Strong consistency (D) is unrelated to throughput cost.

Source: Microsoft Learn — Optimize provisioned throughput cost →
Question 9 · Maintain (25–30%)

Security policy requires that application code and administrators authenticate to Cosmos DB using Microsoft Entra ID identities and role-based access control rather than primary account keys.

Which approach meets this requirement?

  1. Share the primary key across all applications
  2. Disable all authentication in development
  3. Store the primary key in application source code
  4. Use Microsoft Entra ID with Azure RBAC role assignments for data-plane access
Show answer & explanation

Correct: D — use Microsoft Entra ID with Azure RBAC role assignments for data-plane access. Entra ID identities with RBAC role assignments grant least-privilege, auditable access without distributing account keys — exactly the key-free model the policy requires.

Why not the others: sharing the primary key (A) and embedding it in source (C) are exactly the key-based patterns the policy forbids; disabling authentication (B) is insecure and not a real option.

Source: Microsoft Learn — Entra ID RBAC for Cosmos DB → Further reading: PowerKram — AZ-204 Azure Developer Associate →
Question 10 · Maintain (25–30%)

A team needs the ability to restore a Cosmos DB container to any second within the retention window after an accidental bulk delete, rather than to fixed periodic snapshots.

Which backup option should you configure?

  1. Continuous backup with point-in-time restore
  2. Periodic backup only
  3. A composite index
  4. The analytical store
Show answer & explanation

Correct: A — continuous backup with point-in-time restore. Continuous backup lets you restore to any point in time within the retention window, which is what recovering to the moment just before an accidental bulk delete requires — exactly the scenario.

Why not the others: periodic backup (B) only restores to fixed snapshot intervals, not any second; a composite index (C) speeds queries; the analytical store (D) supports analytics, not restore.

Source: Microsoft Learn — Backup and restore →

Keep going: study & career resources

DP-420 is a developer Specialty on the Azure data track. Below it, the data-fundamentals exam builds the base — and the data-engineer career track shows where these Cosmos DB and application skills lead.

Deep dive: DP-420 format, the five areas, the Specialty tier, and study path

Exam format and scoring

DP-420 delivers roughly 40 to 60 questions in about 100 to 120 minutes, with a passing score of 700 out of 1000 (scaled). It includes multiple choice, multiple response, drag-and-drop, and one or more case studies. It is a developer exam: expect code-adjacent scenarios about partition keys, SQL queries, SDK operations, consistency, and RU cost — often requiring you to read C# or Java and reason about JSON. There is no penalty for wrong answers. See the AZ-204 Azure Developer path →

Why data modeling dominates

At 35–40%, data modeling is by far the heaviest area — and partitioning is its most-tested topic. Choosing a partition key that spreads storage and throughput evenly while matching access patterns, and knowing when to use synthetic or hierarchical keys, is the single most valuable skill for this exam. The SDK data-access sub-topics (point vs query operations, transactional batch, bulk, ETags, pagination, 429 handling) and server-side JavaScript also live in this area. See the DP-900 Data Fundamentals path →

The Specialty tier

Azure Cosmos DB Developer Specialty is a focused, single-technology certification rather than a broad associate credential. It goes deep on one service, so the questions assume you are actually building on Cosmos DB, not just aware of it. Like associate and expert certifications, Specialty credentials are valid for one year and renew free through an online assessment on Microsoft Learn. See the AZ-900 Azure Fundamentals path →

Request units, consistency, and cost

Cosmos DB bills throughput in request units (RU/s), and much of the DP-420 optimization and maintenance content is about controlling RU cost: right-sizing throughput (serverless vs provisioned vs autoscale), tuning indexing policies for read- versus write-heavy workloads, and choosing a consistency level that balances correctness against latency and RU cost. Understanding how each choice moves the cost and performance needle is a recurring theme. See the DP-900 Data Fundamentals path →

Realistic study path

Most candidates come to DP-420 with hands-on Azure development experience and need four to eight weeks of focused study. Work through the official Microsoft Learn DP-420 learning path, build a real Cosmos DB for NoSQL app with the C# SDK (partitioning experiments, transactional batch, change feed with Azure Functions, custom indexing, RBAC), then drill scenario questions in PowerKram’s Learn mode and finish with timed Exam-mode runs. Spend the most time on data modeling and maintenance, since together they are roughly two-thirds of the exam. See the AZ-204 Azure Developer path →

Career outlook and next steps

DP-420 signals deep, provable expertise in a specific high-value technology, which stands out for application-developer, cloud-native-engineer, and data-platform-developer roles building globally distributed apps. It pairs naturally with AZ-204 for broader Azure development and builds on DP-900 fundamentals. For SQL and NoSQL developers moving to cloud-native architectures, it is a strong differentiator. Career Hub — Data Engineer →

Microsoft DP-420 (Azure Cosmos DB Developer Specialty) exam FAQ

What are the DP-420 skill areas and weights?
Per Microsoft’s current DP-420 study guide (skills measured as of April 20, 2026): Design and implement data models (35–40%), Design and implement data distribution (5–10%), Integrate an Azure Cosmos DB solution (5–10%), Optimize an Azure Cosmos DB solution (15–20%), and Maintain an Azure Cosmos DB solution (25–30%). Data modeling is by far the heaviest area, and maintenance is a strong second. Microsoft publishes these as ranges.
Is DP-420 still active?
Yes. DP-420 is a current, active exam and the single required exam for the Microsoft Certified: Azure Cosmos DB Developer Specialty certification. Its blueprint was last updated April 20, 2026, with only minor changes, and there is no announced retirement date.
Is DP-420 a developer or an administrator exam?
It is a developer exam. Microsoft expects you to design data models, write efficient SQL queries and indexing policies, use the C# or Java SDKs, write server-side JavaScript (stored procedures, triggers, UDFs), and reason about consistency and request-unit cost. It is a Specialty credential focused specifically on building applications with Azure Cosmos DB for NoSQL.
What is the DP-420 passing score and format?
A passing score of 700 out of 1000 (scaled), with roughly 40 to 60 questions in about 100 to 120 minutes, including one or more case studies. Formats include multiple choice, multiple response, and drag-and-drop. There is no penalty for incorrect answers.
Does the certification expire?
Yes. Microsoft specialty certifications like Azure Cosmos DB Developer Specialty are valid for one year and must be renewed. Renewal is free through an online assessment on Microsoft Learn during the six months before the certification expires.

Start your free 24-hour DP-420 practice trial

Full access to the question bank, both study modes, source-linked explanations, and score-by-skill-area. No credit card required.

Start free trial →