Question bank
Azure Data Engineer Associate
40 Azure Data Engineer Associate questions with the correct answer and a written explanation, organised across 3 exam domains.
- 40 questions
- 3 domains
- Foundation level
certilumQuestion index
Jump straight to any question. Answers and explanations are inline.
Design and implement data storage
8 questions in this domain
- Question 1Single Choice
Which misconception can lead to query performance degradation when partitioning tables in a data warehouse?
- AMore partitions always improve query performanceCorrect answer
- BUse a non-numeric partition key
- CFact tables should never be partitioned
- DOnly small tables need partitioning
Why this is the answer
Over-partitioning creates many small partitions and extra overhead. The goal is to improve partition elimination, not to maximize the number of partitions.
- Question 2Multiple Select
Which statements about using CSV instead of optimized columnar formats for an enterprise analytical lake are correct?
- ACSV supports columnar compression
- BParquet uses columnar storage for analyticsCorrect answer
- CCSV is universally the best format for analytical lakes
- DFile format affects query performance and storage costCorrect answer
Why this is the answer
Columnar formats such as Parquet improve scan performance and compression. CSV lacks these optimizations, and file-format choice significantly affects cost and query performance.
- Question 3Interactive
Review the linked service definition in the code block. What security issue is present in this configuration?
- AHardcoded database password in the JSON definitionCorrect answer
- BMissing integration runtime
- CIncorrect server name format
- DMissing database name
Why this is the answer
Hardcoding secrets such as passwords in linked-service JSON exposes credentials. Use Azure Key Vault or another secure secret store for database passwords.
- Question 4Single Choice
Which component is required for Azure Data Factory to retrieve data from an on-premises SQL Server?
- ASelf-hosted integration runtimeCorrect answer
- BAzure Integration Runtime
- CAzure Databricks cluster
- DApache Spark pool
Why this is the answer
A self-hosted integration runtime is installed on-premises or in a virtual network to securely connect Azure Data Factory to on-premises SQL Server.
- Question 5Multiple Select
Which actions help configure high availability for a self-hosted integration runtime in Azure Data Factory?
- AInstall on multiple nodes and register them togetherCorrect answer
- BUse only one node to reduce complexity
- CRegister all nodes with the same integration runtimeCorrect answer
- DStore credentials in the browser
Why this is the answer
High availability is achieved by installing and registering multiple nodes in the same self-hosted integration runtime, preventing a single point of failure.
- Question 6Interactive
Examine the Azure Cosmos DB indexing policy in the code block. What should be considered before adding many more included paths?
- ARequest Unit consumptionCorrect answer
- BVirtual network latency
- CData warehouse distribution
- DIntegration runtime capacity
Why this is the answer
Indexes consume storage and Request Units per operation. Analyzing RU impact prevents indiscriminate indexing that hurts performance and increases cost.
- Question 7Single Choice
Which file format is most commonly recommended for large enterprise analytical lakes in Azure Data Lake Storage?
- AParquetCorrect answer
- BJSON
- CCSV
- DXML
Why this is the answer
Parquet is a columnar format that improves read performance and compression. CSV and JSON are less efficient for large analytics workloads.
- Question 8Multiple Select
Which factors should be analyzed before applying secondary indexes in Azure Cosmos DB containers?
- ARequest Unit cost per operationCorrect answer
- BIndex storage consumptionCorrect answer
- CAccount region
- DNumber of virtual machines
Why this is the answer
Additional indexes increase Request Unit charges and storage footprint. Evaluating these costs prevents indiscriminate index creation in Cosmos DB.
Develop data processing
16 questions in this domain
- Question 9Single Choice
A data engineer assumes that adding more partitions will always improve query performance. What is the correct understanding of over-partitioning in data processing?
- AIt always reduces query time.
- BIt can introduce overhead and degrade performance.Correct answer
- CIt has no effect on query performance.
- DIt is the recommended way to scale any workload.
Why this is the answer
Over-partitioning adds scheduling and metadata overhead that can degrade performance instead of improving it, according to the common traps.
- Question 10Multiple Select
Which of the following are identified as common traps when developing data processing solutions? (Select all that apply.)
- AAssuming more partitions always improves query performance.Correct answer
- BUsing CSV formats for large enterprise analytical lakes.Correct answer
- CHardcoding passwords or secrets in linked service JSON.Correct answer
- DUsing Azure Key Vault to store credentials.
Why this is the answer
The listed traps include over-partitioning, CSV-only analytics, and hardcoded secrets. Azure Key Vault is a secure alternative and not a trap.
- Question 11Interactive
Review the Azure Data Factory linked service JSON. Which practice shown in the code should be avoided?
- AUsing a self-hosted integration runtime
- BHardcoding the database password in the JSON definitionCorrect answer
- CEnabling encryption in the connection string
- DSpecifying the server name in the connection string
Why this is the answer
The JSON definition contains a plaintext database password, which is one of the common traps in developing ADF linked services.
- Question 12Single Choice
Which symptom is most likely to appear in a workload that has been over-partitioned?
- AIncreased overhead and degraded performanceCorrect answer
- BFully optimized parallel execution
- CA consistent decrease in resource usage
- DImmediate failure of all queries
Why this is the answer
Over-partitioning produces scheduling and coordination overhead, frequently resulting in degraded performance.
- Question 13Multiple Select
Which statements about adopting CSV formats in large enterprise analytical lakes are consistent with the common traps guidance? (Select all that apply.)
- ACSV is always the most compatible and best choice.
- BCSV may not enforce a schema.Correct answer
- CCSV can cause performance challenges at scale.Correct answer
- DCSV offers better compression than columnar formats.
Why this is the answer
CSV data can lack schema enforcement and often underperforms at scale when used in analytical lakes.
- Question 14Interactive
Review the Azure Data Factory linked service JSON for an on-premises SQL Server. Which configuration is causing connectivity to fail?
- AIt uses the default Azure runtime instead of a Self-Hosted Integration Runtime.Correct answer
- BIt specifies an on-premises server name.
- CIt uses SQL Server authentication.
- DIt includes a database name.
Why this is the answer
Azure Data Factory cannot natively reach on-premises SQL Server over the public internet; a Self-Hosted Integration Runtime is required.
- Question 15Single Choice
What is the effect of over-partitioning on query performance in distributed data processing?
- AIt always improves performance.
- BIt can degrade performance because of task overhead.Correct answer
- CIt only increases storage costs.
- DIt has no measurable effect.
Why this is the answer
The extra tasks and coordination overhead created by over-partitioning can degrade performance.
- Question 16Multiple Select
Which practices are listed as common traps and should be avoided when developing data processing solutions? (Select all that apply.)
- ACreating an excessive number of partitions.Correct answer
- BStoring credentials in Azure Key Vault.
- CHardcoding service principal secrets in linked service JSON.Correct answer
- DAdding secondary indexes without analyzing Request Unit consumption.Correct answer
Why this is the answer
The traps include over-partitioning, hardcoded secrets, and indiscriminate indexing; Key Vault is recommended.
- Question 17Interactive
An engineer reviews a Cosmos DB indexing policy. What key metric should be analyzed before adding secondary indexes as shown?
- ARequest Unit (RU) consumptionCorrect answer
- BNetwork egress
- CStorage account replication
- DVirtual machine size
Why this is the answer
Secondary indexes can increase Request Unit consumption; RU analysis should be performed before adding them.
- Question 18Single Choice
Which practice is a common trap when working with Azure Cosmos DB indexes?
- AApplying secondary indexes without analyzing RU consumptionCorrect answer
- BUsing partition keys in a container
- CEnabling automatic indexing
- DExcluding unused paths from indexing
Why this is the answer
Indiscriminately applying secondary indexes without measuring RU consumption is a common trap.
- Question 19Multiple Select
Which statements accurately describe connectivity between Azure Data Factory and an on-premises SQL Server? (Select all that apply.)
- AData Factory can natively connect over the public internet.
- BA Self-Hosted Integration Runtime must be used.Correct answer
- CThe Self-Hosted Integration Runtime should be configured with high availability.Correct answer
- DNo additional network configuration is required because the Self-Hosted IR is optional.
Why this is the answer
An on-premises SQL Server requires a Self-Hosted Integration Runtime, which should also be made highly available.
- Question 20Interactive
A deployment script creates a Self-Hosted Integration Runtime. What change is needed to make it highly available?
- ARegister additional self-hosted nodes in the integration runtime.Correct answer
- BSwitch to the Azure integration runtime.
- CDisable the network firewall.
- DIncrease the size of the on-premises virtual machine.
Why this is the answer
High availability for a Self-Hosted IR is achieved by installing and registering multiple nodes.
- Question 21Single Choice
What is the main benefit of enabling multiple nodes for a Self-Hosted Integration Runtime?
- AHigh availabilityCorrect answer
- BBetter public internet access
- CSmaller storage footprint
- DEliminates the need for a network
Why this is the answer
Multiple nodes make the Self-Hosted IR highly available and prevent a single point of failure.
- Question 22Multiple Select
Which items are examples of hardcoding secrets in Azure Data Factory linked services? (Select all that apply.)
- APlacing a SQL Server password directly in the connection string.Correct answer
- BStoring a connection string in Azure Key Vault.
- CSetting a service principal key directly in JSON.Correct answer
- DUsing a linked service parameter to reference a secret.
Why this is the answer
The playbook defines hardcoding passwords or service principal secrets in linked service JSON as a common trap.
- Question 23Interactive
Examine the Azure Blob Storage linked service JSON. Which element violates the common traps recommendation?
- AThe service principal keyCorrect answer
- BThe service principal ID
- CThe tenant name
- DThe storage account name
Why this is the answer
The service principal key is hardcoded in the JSON, making the linked service definition unsafe.
- Question 24Single Choice
What is the main risk of putting secrets directly in ADF linked service JSON definitions?
- ASecurity exposureCorrect answer
- BSlower query performance
- CHigher RU consumption
- DLess accurate data
Why this is the answer
Plaintext secrets expose credentials to anyone with access to the JSON and increase security risk.
Secure, monitor, and optimize data storage and data processing
16 questions in this domain
- Question 25Single Choice
What is a common consequence of over-partitioning large data storage systems?
- AImproved query performance in every scenario
- BIncreased metadata overhead and small file problemsCorrect answer
- CGuaranteed lower storage costs
- DZero impact on query execution
Why this is the answer
More partitions do not always improve performance. Over-partitioning can produce many small files and additional metadata overhead.
- Question 26Multiple Select
Which limitations should be considered before using CSV as the main format in a large enterprise analytical lake? Select all that apply.
- ACSV files lack native schema and data type enforcementCorrect answer
- BCSV provides excellent compression for numeric data
- CCSV supports efficient predicate pushdown in columnar engines
- DCSV is generally less efficient than columnar formats like ParquetCorrect answer
Why this is the answer
CSV has no type enforcement and is usually less efficient than columnar formats for analytical queries.
- Question 27Interactive
What is the most significant security issue in the linked service definition shown in the code block?
- AThe server name is not parameterized
- BThe database name is missing
- CThe password is stored directly in the JSON definitionCorrect answer
- DThe connection string uses port 1433
Why this is the answer
Embedding the password directly in the JSON definition exposes the secret to anyone with access to the file; use Azure Key Vault.
- Question 28Single Choice
What is the likely impact of applying secondary indexes indiscriminately in Azure Cosmos DB?
- AIt eliminates all Request Unit consumption
- BIt increases storage and Request Unit costsCorrect answer
- CIt automatically improves every query without trade-offs
- DIt has no effect on write performance
Why this is the answer
Secondary indexes consume storage and can increase RU costs, especially for writes.
- Question 29Multiple Select
What is needed when Azure Data Factory must connect to an on-premises SQL Server? Select all that apply.
- AA Self-Hosted Integration Runtime installed in the on-premises environmentCorrect answer
- BA direct connection over the public internet
- CA registered Integration Runtime node with access to the SQL Server networkCorrect answer
- DNo connectivity component if the SQL Server is publicly exposed
Why this is the answer
ADF uses a Self-Hosted Integration Runtime to bridge to databases that are not publicly reachable.
- Question 30Interactive
The integration runtime configuration in the code block is for production. What is the primary concern?
- AIt uses the wrong runtime type
- BIt uses only one node, creating a single point of failureCorrect answer
- CIt is not registered to a data factory
- DIt cannot be used with SQL Server
Why this is the answer
A single-node Self-Hosted Integration Runtime is not highly available; production deployments should use multiple nodes.
- Question 31Single Choice
Which statement about adding more partitions to a data processing system is true?
- AMore partitions guarantee faster queries
- BAt some point, extra partitions create overhead that can hurt performanceCorrect answer
- CPartitioning only affects cost, never performance
- DPartition count should be as high as possible for all workloads
Why this is the answer
There is a trade-off; too many partitions can create overhead and degrade performance.
- Question 32Multiple Select
Why are columnar formats such as Parquet often preferred over CSV in enterprise analytical lakes? Select all that apply.
- ABetter compression due to columnar encodingCorrect answer
- BNative support for schema and data typesCorrect answer
- CHuman readability of raw data
- DRequires no partitioning strategy
Why this is the answer
Parquet is columnar, compresses well, and preserves schema information for analytical processing.
- Question 33Interactive
Refer to the Cosmos DB indexing policy in the code block. What should be evaluated to avoid unnecessary cost?
- AThe policy is invalid because Cosmos DB does not support secondary indexes
- BIndexing every path may increase Request Unit and storage costsCorrect answer
- CSetting indexingMode to consistent prevents indexing
- DThe container must be recreated to change indexing
Why this is the answer
Broad indexing policies index more data, which can increase RU consumption and storage.
- Question 34Single Choice
Where should Azure Data Factory store database passwords to avoid hardcoded secrets in linked service JSON?
- AAzure Key VaultCorrect answer
- BLinked service description
- CDefault values in parameters
- DPipeline variable logs
Why this is the answer
Azure Key Vault centralizes secrets and avoids embedding plaintext credentials in JSON.
- Question 35Multiple Select
Which actions improve high availability for a Self-Hosted Integration Runtime? Select all that apply.
- AInstall and register at least two integration runtime nodesCorrect answer
- BKeep all nodes on the same single machine
- CUse multiple nodes to distribute loadCorrect answer
- DDisable automatic updates
Why this is the answer
Multiple registered nodes on separate machines provide high availability and load distribution.
- Question 36Interactive
Refer to the linked service JSON in the code block. Why is a Self-Hosted Integration Runtime required in this configuration?
- AThe SQL Server is outside Azure and is not directly reachable through the public internetCorrect answer
- BAzure Data Factory always requires Self-Hosted IR for Azure SQL Database
- CThe database name contains a dot
- DSelf-Hosted IR is used only for scheduled triggers
Why this is the answer
A Self-Hosted IR is used when the data store is in a private or on-premises network.
- Question 37Single Choice
In Azure Cosmos DB, how do secondary indexes affect Request Unit consumption?
- AIndexes lower RU consumption for all operations
- BUnnecessary indexes can increase RU consumption and storage costsCorrect answer
- CRU consumption is not affected by indexing policy
- DOnly queries consume RUs; writes never consume RUs
Why this is the answer
Each secondary index adds storage and RU costs, so they should be applied selectively.
- Question 38Multiple Select
Which practices help avoid the over-partitioning trap? Select all that apply.
- AConsider data size and query patternsCorrect answer
- BUse the largest possible partition count
- CMonitor query performance and adjustCorrect answer
- DCreate one partition per row
Why this is the answer
Partition count should be based on data size and queries, then monitored and adjusted.
- Question 39Interactive
A data engineer creates a dataset using the file format configuration in the code block. What is a common consequence in an enterprise analytical lake?
- ACSV provides automatic schema enforcement
- BCSV can create performance and type-safety challengesCorrect answer
- CCSV is optimized for columnar processing
- DCSV files cannot be read by Azure services
Why this is the answer
CSV lacks schema enforcement and is generally less efficient for analytics than columnar formats.
- Question 40Single Choice
Why can Azure Data Factory not use the public internet to connect directly to most on-premises SQL Servers?
- AThe on-premises SQL Server is usually in a private network without a public endpointCorrect answer
- BAzure Data Factory cannot connect to SQL Server at all
- CPublic internet access is always faster
- DIntegration runtimes are not needed
Why this is the answer
Most on-premises SQL Servers are behind firewalls, so the Self-Hosted IR must bridge the connection.
Frequently asked questions
Everything about the Azure Data Engineer Associate question bank and how to use it.