AZ-305 · Domain 2 of 4

AZ-305 Domain 2: Design data storage solutions

This domain covers designing the data layer: choosing relational stores and their tiers and scaling, selecting stores for semi-structured and unstructured data, and designing data integration and analytics pipelines.

Exam format
20–25%
Official functional groups
3
Assessment estimate
9–14
Suggested study time estimate
20–30 hrs

About 10 min read Reviewed October 2026

Overview

Design data storage solutions covers recommending the right data stores and data movement for a workload. It has three objective groups. Relational data covers recommending a relational store such as Azure SQL Database or SQL Managed Instance, the service and compute tier, scalability, and data protection. Semi-structured and unstructured data covers recommending stores such as Cosmos DB, Blob Storage and Data Lake, balancing features, performance and cost, and designing for protection and durability. Data integration covers recommending solutions for data integration and analysis, such as Azure Data Factory and analytics services.

At 20–25%, this domain tests data-platform design judgment: matching the data model, scale, consistency, performance and cost requirements to the right Azure data service. The exam rewards distinguishing relational from non-relational stores, choosing tiers and scaling approaches, and designing integration pipelines, all driven by the stated requirements.

How it's tested

AZ-305 questions in this domain describe a data model, scale, consistency and cost requirement and ask you to recommend the data store or integration design. AZ-305 questions are requirement-driven design scenarios where several options may work, so read every requirement and recommend the design that meets all of them before answering.

  • Relational store and tier

    A relational need is described, and you recommend Azure SQL Database or Managed Instance, the tier and scaling.

  • Non-relational store

    A semi-structured or unstructured need is described, and you recommend Cosmos DB, Blob Storage or Data Lake.

  • Integration pipeline

    A data-movement or analytics need is described, and you recommend Data Factory or an analytics service.

Reasoning it through

An application needs a globally distributed database with single-digit-millisecond latency and flexible schema for a high-throughput key-value workload across regions. Azure SQL Database is relational with a fixed schema and is not designed for global multi-region write distribution at that scale. Azure Cosmos DB provides global distribution, low latency and a flexible schema with multiple consistency levels, which matches the requirement, so Cosmos DB is the recommended store.

Functional groups 3

Official functional groups from the Microsoft outline, in outline order.

2.1 Design data storage solutions for relational dataRelational store selection, service and compute tiers, scalability and protection.
What it means
This covers recommending a relational data store such as Azure SQL Database, Azure SQL Managed Instance or SQL on a VM, recommending the service and compute tier, recommending scalability approaches such as read replicas, elastic pools and hyperscale, and recommending data protection such as backups and encryption.
What to understand
Azure SQL Database is a fully managed PaaS database, Managed Instance offers near full SQL Server compatibility for migration, and SQL on a VM gives full control. Tiers trade performance and cost. Scalability uses elastic pools, read replicas and hyperscale. Data protection includes automated backups, geo-replication and Transparent Data Encryption.
How it shows up
A relational requirement is described, and you must recommend the store, tier, scaling or protection.
Common confusion
Confusing Azure SQL Database with SQL Managed Instance; Managed Instance offers higher SQL Server compatibility, suited to lift-and-shift migration.
2.2 Design data storage solutions for semi-structured and unstructured dataCosmos DB, Blob Storage and Data Lake, balancing features, performance, cost and durability.
What it means
This covers recommending a store for semi-structured data such as Azure Cosmos DB, a store for unstructured data such as Azure Blob Storage or Data Lake Storage, balancing features, performance and cost, and recommending protection and durability with redundancy, tiers and lifecycle policies.
What to understand
Cosmos DB is a globally distributed, multi-model NoSQL database with tunable consistency. Blob Storage holds unstructured objects with access tiers and redundancy options; Data Lake adds hierarchical namespace for analytics. The design balances performance, cost and durability, choosing the right tier and redundancy.
How it shows up
A semi-structured or unstructured requirement is described, and you must recommend the store, tier and durability.
Common confusion
Recommending a relational database for a flexible-schema, globally distributed workload that Cosmos DB serves better.
2.3 Design data integrationRecommend data integration and data analysis solutions.
What it means
This covers recommending a solution for data integration, such as Azure Data Factory for extract, transform and load pipelines, and a solution for data analysis, such as Azure Synapse Analytics or other analytics services.
What to understand
Data integration moves and transforms data between stores, with Azure Data Factory orchestrating pipelines. Data analysis aggregates and queries data at scale with services such as Synapse Analytics. The design recommends how data flows from sources to an analytics destination.
How it shows up
A data-movement or analytics requirement is described, and you must recommend Data Factory or an analytics service.
Common confusion
Confusing data integration (moving and transforming data) with data analysis (querying and aggregating it).

High-yield concepts

The terms, rules and formulas this domain turns into questions most often.

  • Azure SQL Database vs. Managed Instance

    PaaS database versus near full SQL Server compatibility.

    Remember
    Managed Instance suits lift-and-shift migration.
    Tested as
    Relational design questions.
    Trap
    Confusing the two.

  • Service and compute tiers

    Performance and cost levels for Azure SQL.

    Remember
    Match the tier to the workload.
    Tested as
    Relational design questions.
    Trap
    Over- or under-provisioning the tier.

  • Elastic pools and hyperscale

    Shared capacity and massive scale-out options.

    Remember
    Elastic pools share resources; hyperscale scales storage hugely.
    Tested as
    Scalability questions.
    Trap
    Confusing the scaling options.

  • Read replicas

    Scale read traffic and enable failover.

    Remember
    They offload reads.
    Tested as
    Scalability questions.
    Trap
    Expecting them to scale writes.

  • Azure Cosmos DB

    Globally distributed multi-model NoSQL.

    Remember
    Tunable consistency, low latency.
    Tested as
    Non-relational design questions.
    Trap
    Using relational for its use cases.

  • Consistency levels

    Trade-offs from strong to eventual in Cosmos DB.

    Remember
    Stronger consistency costs latency.
    Tested as
    Cosmos DB questions.
    Trap
    Choosing strong where eventual suffices.

  • Blob Storage tiers

    Hot, cool, cold and archive by access frequency.

    Remember
    Match the tier to access patterns.
    Tested as
    Unstructured data questions.
    Trap
    Keeping cold data in hot.

  • Redundancy options

    LRS, ZRS, GRS and variants for durability.

    Remember
    More redundancy, more durability and cost.
    Tested as
    Durability questions.
    Trap
    Confusing redundancy with tiers.

  • Data Lake Storage

    Blob Storage with hierarchical namespace for analytics.

    Remember
    Suited to big-data analytics.
    Tested as
    Analytics questions.
    Trap
    Using plain blob for analytics at scale.

  • Azure Data Factory

    Orchestrates data integration pipelines.

    Remember
    It moves and transforms data.
    Tested as
    Integration questions.
    Trap
    Confusing it with analytics.

  • Azure Synapse Analytics

    Analytics platform for querying data at scale.

    Remember
    It aggregates and analyzes data.
    Tested as
    Analytics questions.
    Trap
    Confusing it with integration.

Side by side

Concepts candidates mix up, compared on the point that decides the answer.

Azure SQL Database vs. Managed Instance vs. SQL on a VM

ConceptMeaningKey differenceCommon trap
Azure SQL Database Fully managed PaaS Lowest management Expecting full SQL Server surface
SQL Managed Instance Near full SQL Server compatibility Best for migration Confusing with SQL Database
SQL on a VM Full control, IaaS Most control, most management Using it where PaaS fits

Relational vs. Cosmos DB vs. Blob/Data Lake

ConceptMeaningKey differenceCommon trap
Relational (Azure SQL) Structured, schema-fixed Strong consistency and joins Using it for global key-value scale
Cosmos DB Globally distributed NoSQL Low latency, flexible schema Expecting relational joins
Blob / Data Lake Unstructured objects Low cost at scale Using it as a transactional store

Data integration vs. data analysis

ConceptMeaningKey differenceCommon trap
Data integration (Data Factory) Moves and transforms data ETL pipelines Confusing with analysis
Data analysis (Synapse) Queries and aggregates data Analytics at scale Confusing with integration

Exam traps

Each trap is an answer that looks right. Learn the clue that should change your mind.

  1. Relational for global NoSQL

    What candidates pick
    Recommend Azure SQL Database for a globally distributed, flexible-schema, high-throughput workload.
    Why it looks right
    Relational databases are familiar.
    Why it's wrong
    Cosmos DB provides global distribution, low latency and flexible schema at scale, which relational does not.

    Clue to watch for The workload is global, high-throughput and flexible-schema.

  2. SQL Database for migration compat

    What candidates pick
    Recommend Azure SQL Database for a lift-and-shift of a SQL Server app needing full compatibility.
    Why it looks right
    It is the managed SQL option.
    Why it's wrong
    SQL Managed Instance offers near full SQL Server compatibility, better for lift-and-shift migration.

    Clue to watch for The migration needs high SQL Server feature compatibility.

  3. Hot tier for archives

    What candidates pick
    Recommend the hot tier for rarely accessed archival data.
    Why it looks right
    Hot is the default.
    Why it's wrong
    Cool, cold or archive tiers cost less for infrequent access; lifecycle policies can move data automatically.

    Clue to watch for The data is accessed rarely.

  4. Strong consistency always

    What candidates pick
    Recommend strong consistency for every Cosmos DB workload.
    Why it looks right
    It guarantees the latest data.
    Why it's wrong
    Strong consistency adds latency and cost; weaker levels suffice for many workloads.

    Clue to watch for The workload tolerates slightly stale reads and latency matters.

  5. Integration is analysis

    What candidates pick
    Recommend Azure Data Factory to run analytical queries over data.
    Why it looks right
    Both handle data.
    Why it's wrong
    Data Factory integrates and moves data; analytics services such as Synapse query and aggregate it.

    Clue to watch for The need is querying and aggregating versus moving data.

Answer framework

For any Domain 2 design scenario, use these steps to find the answer AZ-305 expects.

  1. Identify the data model, scale, consistency, performance, durability and cost requirements before recommending a store.
  2. For relational data, recommend Azure SQL Database, Managed Instance for high compatibility or SQL on a VM for full control, with the right tier and scaling.
  3. For semi-structured or unstructured data, recommend Cosmos DB, Blob Storage or Data Lake, choosing the tier and redundancy for cost and durability.
  4. For data movement, recommend Azure Data Factory for integration and an analytics service such as Synapse for analysis.
  5. Balance features, performance and cost, and match consistency and redundancy to the requirements. Confirm the recommended design meets every stated requirement, including compliance and cost, before committing to it.

Practice question

An application requires a database that is globally distributed across multiple regions, offers single-digit-millisecond read and write latency, and uses a flexible, schema-less document model for a high-throughput workload. Which Azure data store should the architect recommend?

Want more like this? The AZ-305 practice test has questions for every domain.

Checklist

0 of 10

Tick each one you can do without notes. Your ticks stay in this browser.

Study strategy

Approach this domain as matching data requirements to the right Azure service. Start with relational design, learning when to recommend Azure SQL Database, Managed Instance (for high-compatibility migration) or SQL on a VM, and the tier and scaling options such as elastic pools, read replicas and hyperscale.

Next, learn non-relational design: Cosmos DB for globally distributed flexible-schema workloads with tunable consistency, and Blob Storage and Data Lake for unstructured data with the right tier and redundancy. Drill the relational-versus-Cosmos DB choice, which is a frequent decision point. Then learn data integration with Data Factory versus analysis with Synapse.

Because questions are requirement-driven, practice reading the data model, scale, consistency and cost requirements and recommending the store that fits, rather than defaulting to a familiar one. Afterwards, move to Domain 3, designing business continuity solutions. Because AZ-305 is a design exam, drill requirement-driven scenarios and justify why the recommended design beats the alternatives, anchoring on the Well-Architected and Cloud Adoption Frameworks, and review missed questions by group.

FAQ

How much of AZ-305 is Domain 2?

Design data storage solutions is 20–25% of the exam.

What is the difference between Azure SQL Database and SQL Managed Instance?

Azure SQL Database is a fully managed PaaS database. SQL Managed Instance offers near full SQL Server compatibility, making it better suited to lift-and-shift migrations of existing SQL Server workloads.

When should I recommend Cosmos DB?

For globally distributed, low-latency, high-throughput workloads with a flexible or non-relational schema, where a traditional relational database would not scale or fit the model.

How do I choose a Blob Storage tier?

By access frequency: hot for frequent access, cool or cold for infrequent, and archive for rarely accessed data, using lifecycle policies to move data automatically as it cools.

What is the difference between data integration and data analysis?

Data integration, with Azure Data Factory, moves and transforms data between stores. Data analysis, with services such as Synapse Analytics, queries and aggregates data at scale.

What controls cost and latency in Cosmos DB?

The chosen consistency level and throughput. Stronger consistency increases latency and cost, so the design should use the weakest consistency that meets the requirements.

How does Domain 2 connect to the rest of the exam?

The data stores designed here are secured by Domain 1's identity and governance, protected by Domain 3's continuity design, and consumed by Domain 4's infrastructure and application architecture.

Other Cloud Computing certifications

Where the same subject shows up in other exams. Cards marked "Same subject" open the matching domain.

Test your Domain 2 knowledge

You have covered relational, non-relational and integration data design. Test whether you can recommend the data store that meets every requirement under exam conditions.

Take the AZ-305 practice test Check your AZ-305 readiness

Keep going