Databricks-Certified-Professional-Data-Engineer Sample Questions Answers

Delta Lake outperforms legacy Hadoop formats because it leverages Parquet-based storage , data skipping , and file pruning . According to Databricks documentation, Delta Lake automatically stores detailed statistics (min/max values and file-level metadata) in the transaction log. During query planning, the engine uses these statistics to skip entire files that do not match query filters , a process called data skipping and file pruning . Additionally, Delta uses a vectorized Parquet reader , which reduces I/O and CPU overhead. Together, these optimizations allow Delta to scan significantly less data and produce more efficient physical query plans (e.g., Shuffle Hash Join instead of Sort Merge Join). The performance gain is due to efficient data skipping, not the inherent superiority of join type.

This is the correct answer because it describes a limitation of Databricks Secrets. Databricks Secrets is a module that provides tools to store sensitive credentials and avoid accidentally displaying them in plain text. Databricks Secrets allows creating secret scopes, which are collections of secrets that can be accessed by users or groups. Databricks Secrets also allows creating and managing secrets using the Databricks CLI or the Databricks REST API. However, a limitation of Databricks Secrets is that the Databricks REST API can be used to list secrets in plain text if the personal access token has proper credentials. Therefore, users should still be careful with which credentials are stored in Databricks Secrets and which users have access to using these secrets. Verified References: [Databricks Certified Data Engineer Professional], under “Databricks Workspace” section; Databricks Documentation, under “List secrets” section.

Exact extract: “window(timeColumn, windowDuration[, slideDuration]) returns a time window for grouping on event-time.”

Exact extract: “If slideDuration is not specified the windows are non-overlapping (tumbling) with length windowDuration.”

Given the CASE statement in the view definition, the result set for a user not in the auditing group would be constrained by the ELSE condition, which filters out records based on age. Therefore, the view will return all columns normally for records with an age greater than 18, as users who are not in the auditing group will not satisfy the is_member( ' auditing ' ) condition. Records not meeting the age > 18 condition will not be displayed.

Cmd 6 is the command that should be removed from the notebook before scheduling it as a job. This command is selecting all the columns from the finalDF dataframe and displaying them in the notebook. This is not necessary for the job, as the finalDF dataframe is already written to a table in Cmd 7. Displaying the dataframe in the notebook will only consume resources and time, and it will not affect the output of the job. Therefore, Cmd 6 is redundant and should be removed.

The other commands are essential for the job, as they perform the following tasks:

Cmd 1: Reads the raw_data table into a Spark dataframe called rawDF.

Cmd 2: Prints the schema of the rawDF dataframe, which is useful for debugging and understanding the data structure.

Cmd 3: Selects all the columns from the rawDF dataframe, as well as the nested columns from the values struct column, and creates a new dataframe called flattenedDF.

Cmd 4: Drops the values column from the flattenedDF dataframe, as it is no longer needed after flattening, and creates a new dataframe called finalDF.

Cmd 5: Explains the physical plan of the finalDF dataframe, which is useful for optimizing and tuning the performance of the job.

Cmd 7: Writes the finalDF dataframe to a table called flat_data, using the append mode to add new data to the existing table.

This is the correct answer because it meets the requirements of maintaining a full record of all values that have ever been valid in the source system and recreating the current table state with only the most recent value for each record. The code ingests all log information into a bronze table, which preserves the raw CDC data as it is. Then, it uses merge into to perform an upsert operation on a silver table, which means it will insert new records or update or delete existing records based on the change type and the pk_id columns. This way, the silver table will always reflect the current state of the source table, while the bronze table will keep the history of all changes. Verified References: [Databricks Certified Data Engineer Professional], under “Delta Lake” section; Databricks Documentation, under “Upsert into a table using merge” section.

The Databricks documentation specifies that for Lakeflow Declarative Pipelines, detailed data quality metrics are logged as events of type expectation_result within the event log. Each record of this type contains fields including expectation_name, dataset_name, passed_records, and failed_records. Filtering on event_type = ' expectation_result ' and expanding the details field allows retrieving metrics for each expectation from the most recent pipeline update. While flow_progress provides summary statistics and data_quality events aggregate results, only expectation_result events provide granular, per-expectation metrics required for audit and monitoring automation.

This is the correct answer because stream-static joins are supported by Structured Streaming when one of the tables is a static Delta table. A static Delta table is a Delta table that is not updated by any concurrent writes, such as appends or merges, during the execution of a streaming query. In this case, each microbatch of a stream-static join will use the most recent version of the static Delta table as of each microbatch, which means it will reflect any changes made to the static Delta table before the start of each microbatch. Verified References: [Databricks Certified Data Engineer Professional], under “Structured Streaming” section; Databricks Documentation, under “Stream and static joins” section.

Databricks provides different permission levels to control access to clusters. The correct minimal permission required for viewing driver logs is CAN VIEW .

Databricks Cluster Permission Levels:

CAN ATTACH TO:

Allows users to attach notebooks to a cluster but does not allow them to view logs .

Not sufficient for viewing driver logs.

CAN MANAGE:

Grants full control over the cluster, including starting, stopping, and editing configurations.

Too broad for this requirement .

CAN VIEW (Correct Answer):

Allows users to view cluster details, logs, and status but not modify any configurations.

Minimal required permission for viewing logs .

CAN RESTART:

Grants permission to restart the cluster , but does not include log access .

Not sufficient for viewing logs.

Conclusion:

The minimal permission needed to allow the development team to view driver logs is CAN VIEW .

In Unity Catalog, privileges are non-cascading —meaning that granting a privilege (like SELECT) on a catalog does not automatically grant the same privilege on contained objects (schemas, tables, or views). Each object type has its own independent access control hierarchy.

According to the Databricks access control documentation: “Privileges do not automatically cascade from catalog to schema or table levels.” Administrators must explicitly grant privileges on each level if users need access across objects. This design ensures tighter governance and least-privilege enforcement. Therefore, option B correctly describes Unity Catalog’s privilege model, while A and D incorrectly imply automatic inheritance.

Exact extract: “select() expects column names or Column expressions.”

Exact extract: “When using strings directly, Spark SQL interprets them as literal column names.”

Exact extract: “Python string operations, such as " colname " *3, return repeated strings, not column expressions.”

The expression 3* " heartrate " is Python string multiplication, which evaluates to " heartrateheartrateheartrate " . The select() method interprets this as a literal column name. Since there is no column with that name in the DataFrame schema, Spark raises AnalysisException saying it cannot resolve that column. To correctly multiply a column by a scalar, one must use the column expression form:

from pyspark.sql.functions import col

df.select((col( " heartrate " ) * 3).alias( " heartrate_x3 " ))

This ensures Spark evaluates the arithmetic operation on the column instead of misinterpreting the string.

End-to-end testing is a methodology used to test whether the flow of an application, from start to finish, behaves as expected. The key benefit of an end-to-end test is that it closely simulates real-world, user behavior, ensuring that the system as a whole operates correctly.

The correct method to define a table using Lakeflow Declarative Pipelines (LDP) is with the @dlt.table decorator, which persists the output as a managed Delta table. When ingesting raw JSON data, spark.read.json() or spark.read.format( " json " ).load() is the standard approach. This reads JSON-formatted files from the source and stores them in Delta format automatically managed by Databricks.

Reference Source: Databricks Lakeflow Declarative Pipelines Developer Guide – “Create tables from raw JSON and Delta sources.”

This is the correct answer because the code uses the dropDuplicates method to remove any duplicate records within each batch of data before writing to the orders table. However, this method does not check for duplicates across different batches or in the target table, so it is possible that newly written records may have duplicates already present in the target table. To avoid this, a better approach would be to use Delta Lake and perform an upsert operation using mergeInto. Verified References: [Databricks Certified Data Engineer Professional], under “Delta Lake” section; Databricks Documentation, under “DROP DUPLICATES” section.

A unit test is designed to verify the correctness of a small, isolated piece of code, typically a single function. Testing a mathematical function that calculates the area under a curve is an example of a unit test because it is testing a specific, individual function to ensure it operates as expected.

Databricks Catalog Explorer provides a feature called AI Generate that automatically produces intelligent comments for columns. This feature uses metadata such as column names, types, patterns, and sampled values to generate human-readable documentation. According to the documentation, this is the recommended method to rapidly enrich schema metadata and improve data discoverability, especially at enterprise scale. Unlike DESCRIBE HISTORY or DESCRIBE TABLE, which only surface technical schema details, AI Generate directly produces business-oriented descriptions. PySpark statistical functions (df.describe) only return numeric statistics and cannot generate descriptive metadata. Thus, AI Generate in Catalog Explorer is the correct approach.

The Databricks documentation recommends applyInPandas() for complex per-group operations where maintaining internal state within each group is necessary. When using applyInPandas(), Spark provides all records for each grouping key as a Pandas DataFrame to the function, allowing efficient vectorized operations with local state management. This approach ensures high performance and scalability while maintaining logical isolation between groups. In contrast, SCALAR and SCALAR_ITER UDFs operate on individual rows or batches and cannot maintain inter-row state effectively. grouped_agg UDFs are limited to computing aggregates and do not support complex multi-row transformations. Therefore, applyInPandas() is the correct and Databricks-recommended solution for stateful per-group time-series computations.

This is the correct answer because skew is a common situation that causes increased duration of the overall job. Skew occurs when some partitions have more data than others, resulting in uneven distribution of work among tasks and executors. Skew can be caused by various factors, such as skewed data distribution, improper partitioning strategy, or join operations with skewed keys. Skew can lead to performance issues such as long-running tasks, wasted resources, or even task failures due to memory or disk spills. Verified References: [Databricks Certified Data Engineer Professional], under “Performance Tuning” section; Databricks Documentation, under “Skew” section.

This is the correct answer because it is the cluster configuration that will result in maximum performance for a job with at least one wide transformation. A wide transformation is a type of transformation that requires shuffling data across partitions, such as join, groupBy, or orderBy. Shuffling can be expensive and time-consuming, especially if there are too many or too few partitions. Therefore, it is important to choose a cluster configuration that can balance the trade-off between parallelism and network overhead. In this case, having 8 VMs with 50 GB per executor and 20 cores per executor will create 8 partitions, each with enough memory and CPU resources to handle the shuffling efficiently. Having fewer VMs with more memory and cores per executor will create fewer partitions, which will reduce parallelism and increase the size of each shuffle block. Having more VMs with less memory and cores per executor will create more partitions, which will increase parallelism but also increase the network overhead and the number of shuffle files. Verified References: [Databricks Certified Data Engineer Professional] , under “Performance Tuning” section; Databricks Documentation, under “Cluster configurations” section.

 This is the correct answer because it describes how Delta Lake can help to avoid data loss of this nature in the future. By ingesting all raw data and metadata from Kafka to a bronze Delta table, Delta Lake creates a permanent, replayable history of the data state that can be used for recovery or reprocessing in case of errors or omissions in downstream applications or pipelines. Delta Lake also supports schema evolution, which allows adding new columns to existing tables without affecting existing queries or pipelines. Therefore, if a critical field was omitted from an application that writes its Kafka source to Delta Lake, it can be easily added later and the data can be reprocessed from the bronze table without losing any information. Verified References: [Databricks Certified Data Engineer Professional], under “Delta Lake” section; Databricks Documentation, under “Delta Lake core features” section.

The libraries command group allows you to install, uninstall, and list libraries on Databricks clusters. You can use the libraries install command to install a custom Python Wheel on a cluster by specifying the --whl option and the path to the wheel file. For example, you can use the following command to install a custom Python Wheel named mylib-0.1-py3-none-any.whl on a cluster with the id 1234-567890-abcde123:

databricks libraries install --cluster-id 1234-567890-abcde123 --whl dbfs:/mnt/mylib/mylib-0.1-py3-none-any.whl

This will upload the custom Python Wheel to the cluster and make it available for use with a production job. You can also use the libraries uninstall command to uninstall a library from a cluster, and the libraries list command to list the libraries installed on a cluster.

Lakeflow Declarative Pipelines provide the expect_all method for programmatically applying multiple data quality expectations at once. The documentation explains that @dlt.expect_all accepts a dictionary of expectation names mapped to SQL constraints, allowing rules to be dynamically loaded from metadata such as JSON files. This ensures that pipelines remain maintainable and scalable without needing to hardcode individual @dlt.expect decorators. The event logs will track each expectation’s pass and fail counts individually, making it auditable. Other options are incorrect: invoking an external API introduces unnecessary complexity, individual decorators require hardcoding, and SQL constraints cannot dynamically reference external JSON.

To deduplicate data against previously processed records as it is inserted into a Delta table, you can use the merge operation with an insert-only clause. This allows you to insert new records that do not match any existing records based on a unique key, while ignoring duplicate records that match existing records. For example, you can use the following syntax:

MERGE INTO target_table USING source_table ON target_table.unique_key = source_table.unique_key WHEN NOT MATCHED THEN INSERT *

This will insert only the records from the source table that have a unique key that is not present in the target table, and skip the records that have a matching key. This way, you can avoid inserting duplicate records into the Delta table.

The hash() function in Databricks SQL returns a deterministic fixed-length integer (or hexadecimal string) derived from the input. When applied to sensitive identifiers like email addresses, it produces a unique value for each distinct input while ensuring uniform output size, making it suitable for anonymization where referential consistency is required.

Functions like mask() perform pattern-based substitutions that change string lengths, and sha1() or sha2() produce long hexadecimal strings of varying lengths (depending on hash size), which may not match requirements for fixed-length masking.

Therefore, the correct choice for fixed-length, deterministic pseudonymization of email addresses is hash(email) , as it maintains analytical usability while anonymizing sensitive data.

When Spark performs wide transformations such as sortBy or orderBy, large data volumes can exceed memory limits, causing disk spilling . The official Databricks performance tuning guide recommends increasing the shuffle partition count to distribute the data more evenly across executors. By default, Spark uses a fixed number of shuffle partitions (e.g., 200), which can lead to memory imbalance and spill if some partitions are too large. Increasing this number (via spark.sql.shuffle.partitions) results in smaller partitions, reduced in-memory pressure, and improved sort performance. Other options like broadcast joins or single partition sorts do not apply to single-table sorts, and converting to filters changes query logic. Thus, option D is the correct remedy.

The GRANT USAGE ON DATABASE prod TO eng command grants the eng group the permission to use the prod database, which means they can list and access the tables and views in the database. The GRANT SELECT ON DATABASE prod TO eng command grants the eng group the permission to select data from the tables and views in the prod database, which means they can query the data using SQL or DataFrame API. However, these commands do not grant the eng group any other permissions, such as creating, modifying, or deleting tables and views, or defining custom functions. Therefore, the eng group members are able to query all tables and views in the prod database, but cannot create or edit anything in the database. References:

Grant privileges on a database : https://docs.databricks.com/en/security/auth-authz/table-acls/grant-privileges-database.html

Privileges you can grant on Hive metastore objects : https://docs.databricks.com/en/security/auth-authz/table-acls/privileges.html

Exact extract: “Job clusters are created for a job run and terminate when the job completes.”

Exact extract: “select() expects column names or Column expressions.”

Databricks recommends addressing disk spilling —which occurs when Spark tasks run out of memory—by increasing memory per core and controlling partition size. Selecting an instance type with a higher memory-to-core ratio (A) provides each task with more available RAM, directly reducing the chance of spilling to disk. Additionally, reducing spark.sql.files.maxPartitionBytes (D) creates smaller partitions, preventing any single task from holding too much data in memory. Increasing partition size (C) or disk capacity (B) does not solve memory bottlenecks, and bandwidth (E) affects network I/O, not spill behavior. Therefore, the correct actions are A and D .

Converting the data to Delta Lake may not improve query performance on free text fields with high cardinality, such as the review column. This is because Delta Lake collects statistics on the minimum and maximum values of each column, which are not very useful for filtering or skipping data on free text fields. Moreover, Delta Lake collects statistics on the first 32 columns by default, which may not include the review column if the table has more columns. Therefore, the junior data engineer’s suggestion is not correct. A better approach would be to use a full-text search engine, such as Elasticsearch, to index and query the review column. Alternatively, you can use natural language processing techniques, such as tokenization, stemming, and lemmatization, to preprocess the review column and create a new column with normalized terms that can be used for filtering or skipping data. References:

Optimizations : https://docs.delta.io/latest/optimizations-oss.html

Full-text search with Elasticsearch : https://docs.databricks.com/data/data-sources/elasticsearch.html

Natural language processing : https://docs.databricks.com/applications/nlp/index.html

https://docs.databricks.com/en/delta/delta-change-data-feed.html

The issue with the refactored code is that it tries to use string interpolation to dynamically create table names within the dlc.table decorator, which will not correctly interpret the table names. Instead, by using a dictionary with table names as keys and their configurations as values, the data engineer can iterate over the dictionary items and use the keys (table names) to properly configure the table settings. This way, the decorator can correctly recognize each table name, and the corresponding configuration settings can be applied appropriately.

In the Spark UI ' s Storage tab, an indicator that a cached table is not performing optimally would be the presence of the _disk annotation in the RDD Block Name. This annotation indicates that some partitions of the cached data have been spilled to disk because there wasn ' t enough memory to hold them. This is suboptimal because accessing data from disk is much slower than from memory. The goal of caching is to keep data in memory for fast access, and a spill to disk means that this goal is not fully achieved.

Delta Lake introduced Delta Universal Format (Delta UniForm) , which allows seamless interoperability between Delta Lake and Apache Iceberg . This means a Delta table can be converted into an Iceberg table while maintaining Delta capabilities .

Explanation of Each Option:

(A) Require the analytics team to use a tool that supports Delta table

Incorrect: While Delta Lake is widely used, requiring the team to change tools is not a flexible or scalable solution .

(B) Enable uniform on the transactions table to ' iceberg ' so that the table can be read as an Iceberg table

Incorrect:

The uniform feature must be enabled after conversion .

You cannot directly enable uniform without first converting the table .

(C) Create an Iceberg copy of the transactions Delta table which can be used by the analytics team

Incorrect:

Creating a separate Iceberg copy would duplicate storage and increase maintenance complexity .

This is not necessary when Delta UniForm allows direct compatibility with Iceberg .

(D) Convert the transactions Delta table to Iceberg and enable uniform so that the table can be read as a Delta table

Correct:

The best approach is to convert the existing Delta table to Iceberg using the Databricks Delta to Iceberg migration tools .

After conversion, enabling uniform ensures the table remains accessible in both Delta and Iceberg formats .

Conclusion:

The best practice for interoperability between Delta and Iceberg is to convert the Delta table to Iceberg and enable uniform , ensuring cross-compatibility without data duplication.

Thus, Option (D) is the correct answer .

The Query Profiler in Databricks SQL and notebooks provides a detailed breakdown of query performance metrics. The “Total wall-clock duration” metric directly represents the total elapsed time from query start to completion, including all execution, planning, and waiting stages. In contrast, “Aggregated task time” reflects the cumulative duration across all parallel tasks, which does not equal the total elapsed wall time since tasks often run concurrently. Using job duration from Spark UI can underestimate or overestimate runtime when queries span multiple jobs. Therefore, the Query Profiler’s total wall-clock duration is the officially documented method to determine actual query execution time.

Scheduling a job to execute the data processing pipeline once an hour on a new job cluster is the most cost-effective solution given the scenario. Job clusters are ephemeral in nature; they are spun up just before the job execution and terminated upon completion, which means you only incur costs for the time the cluster is active. Since the total processing time is only 10 minutes, a new job cluster created for each hourly execution minimizes the running time and thus the cost, while also fulfilling the requirement for hourly data updates for the business reporting team ' s dashboards.

The best practice for running unit tests on functions that interact with data is to use a dataset that closely mirrors the production data. This approach allows data engineers to validate the logic of their functions without the risk of affecting the actual production data. It ' s important to have a representative sample of production data to catch edge cases and ensure the functions will work correctly when used in a production environment.

When a Databricks job runs multiple tasks with dependencies, the tasks are executed in a dependency graph. If a task fails, the downstream tasks that depend on it are skipped and marked as Upstream failed. However, the failed task may have already committed some changes to the Lakehouse before the failure occurred, and those changes are not rolled back automatically. Therefore, the job run may result in a partial update of the Lakehouse. To avoid this, you can use the  transactional writes  feature of Delta Lake to ensure that the changes are only committed when the entire job run succeeds. Alternatively, you can use the  Run if  condition to configure tasks to run even when some or all of their dependencies have failed, allowing your job to recover from failures and continue running. References:

transactional writes : https://docs.databricks.com/delta/delta-intro.html#transactional-writes

Run if : https://docs.databricks.com/en/workflows/jobs/conditional-tasks.html

https://spark.apache.org/docs/3.1.3/api/python/reference/api/pyspark.sql.functions.broadcast.html

This is the correct answer because it is the maximum notebook permissions that can be granted to the user without allowing accidental changes to production code or data. Notebook permissions are used to control access to notebooks in Databricks workspaces. There are four types of notebook permissions: Can Manage, Can Edit, Can Run, and Can Read. Can Manage allows full control over the notebook, including editing, running, deleting, exporting, and changing permissions. Can Edit allows modifying and running the notebook, but not changing permissions or deleting it. Can Run allows executing commands in an existing cluster attached to the notebook, but not modifying or exporting it. Can Read allows viewing the notebook content, but not running or modifying it. In this case, granting Can Read permission to the user will allow them to review the production logic in the notebook without allowing them to make any changes to it or run any commands that may affect production data. Verified References: [Databricks Certified Data Engineer Professional] , under “Databricks Workspace” section; Databricks Documentation, under “Notebook permissions” section.

In the Databricks Git folders integration, when merge conflicts arise in notebooks, the UI provides a visual diff editor that highlights conflicting code segments. Users can manually choose which changes to keep from each branch, edit directly in the notebook UI, and remove conflict markers.

After resolving, the engineer must mark the conflict as resolved, save, and commit the final version.

This process ensures that both contributors’ valid code segments are merged correctly and version history is maintained.

Forcing a push (C) or deleting notebooks (B) introduces data loss or versioning issues. Aborting without review (A) violates collaborative best practices. Therefore, D is the only correct and Databricks-approved way to resolve notebook merge conflicts.

This is the correct answer because it describes how data will be filtered when a query is run with the following filter: longitude < 20 and longitude > -20. The query is run on a Delta Lake table that has the following schema: user_id LONG, post_text STRING, post_id STRING, longitude FLOAT, latitude FLOAT, post_time TIMESTAMP, date DATE. This table is partitioned by the date column. When a query is run on a partitioned Delta Lake table, Delta Lake uses statistics in the Delta Log to identify data files that might include records in the filtered range. The statistics include information such as min and max values for each column in each data file. By using these statistics, Delta Lake can skip reading data files that do not match the filter condition, which can improve query performance and reduce I/O costs. Verified References: [Databricks Certified Data Engineer Professional], under “Delta Lake” section; Databricks Documentation, under “Data skipping” section.

The correct answer is A. withWatermark(“event_time”, “10 minutes”). This is because the question asks for incremental state information to be maintained for 10 minutes for late-arriving data. The withWatermark method is used to define the watermark for late data. The watermark is a timestamp column and a threshold that tells the system how long to wait for late data. In this case, the watermark is set to 10 minutes. The other options are incorrect because they are not valid methods or syntax for watermarking in Structured Streaming. References:

Watermarking : https://docs.databricks.com/spark/latest/structured-streaming/watermarks.html

Windowed aggregations : https://docs.databricks.com/spark/latest/structured-streaming/window-operations.html

Exact extract: “If no trigger is specified, the default processing-time trigger runs micro-batches as fast as possible.”

Exact extract: “Trigger once processes all available data once and then stops.”

Lakeflow Declarative Pipelines (LDP), formerly Delta Live Tables (DLT), supports enforcing data quality using expectations . Expectations can either:

Track violations (expect) → records that do not meet conditions are flagged but still included in the pipeline.

Drop violations (expect_or_drop) → records that do not meet conditions are excluded from downstream tables.

Fail pipeline on violations (expect_or_fail) → records that fail conditions stop the pipeline.

In this scenario, the requirement explicitly states that invalid records (where customer_id is null or amount ≤ 0) must be dropped . According to the official documentation, the correct method is .expect_or_drop( " expectation_name " , " SQL_predicate " ) applied on the streaming input.

Option A is correct: It uses .expect_or_drop directly within the transformation chain for both rules, ensuring records that fail are removed before writing to the silver table.

Option B incorrectly uses @dlt.expect decorators, which only track violations but do not drop invalid rows.

Option C uses .expect, which also only flags rows, not drop them.

Option D uses @dlt.expect_or_drop decorator syntax, which is not supported in Python API; expect_or_drop must be applied as a method on the DataFrame, not as a decorator.

Therefore, the correct solution is Option A , which ensures compliance by enforcing data quality and dropping invalid rows programmatically during ingestion.

Partitioning a Delta Lake table is a strategy used to improve query performance by dividing the table into distinct segments based on the values of a specific column. This approach allows queries to scan only the relevant partitions, thereby reducing the amount of data read and enhancing performance.

Considerations for Choosing a Partition Column:

Cardinality: Columns with high cardinality (i.e., a large number of unique values) are generally poor choices for partitioning. High cardinality can lead to a large number of small partitions, which can degrade performance.

Query Patterns: The partition column should align with common query filters. If queries frequently filter data based on a particular column, partitioning by that column can be beneficial.

Partition Size: Each partition should ideally contain at least 1 GB of data. This ensures that partitions are neither too small (leading to too many partitions) nor too large (negating the benefits of partitioning).

Evaluation of Columns:

date:

Cardinality: Typically low, especially if data spans over days, months, or years.

Query Patterns: Many analytical queries filter data based on date ranges.

Partition Size: Likely to meet the 1 GB threshold per partition, depending on data volume.

user_id:

Cardinality: High, as each user has a unique ID.

Query Patterns: While some queries might filter by user_id, the high cardinality makes it unsuitable for partitioning.

Partition Size: Partitions could be too small, leading to inefficiencies.

post_id:

Cardinality: Extremely high, with each post having a unique ID.

Query Patterns: Unlikely to be used for filtering large datasets.

Partition Size: Each partition would be very small, resulting in a large number of partitions.

post_time:

Cardinality: High, especially if it includes exact timestamps.

Query Patterns: Queries might filter by time, but the high cardinality poses challenges.

Partition Size: Similar to user_id, partitions could be too small.

Conclusion:

Given the considerations, the date column is the most suitable candidate for partitioning. It has low cardinality, aligns with common query patterns, and is likely to result in appropriately sized partitions.

To validate that all records from the source are included in the derived table, creating a view that performs a left outer join between the validation_copy table and the report table is effective. The view can highlight any discrepancies, such as null values in the report table ' s key columns, indicating missing records. This view can then be referenced in DLT (Delta Live Tables) expectations for the report table to ensure data integrity. This approach allows for a comprehensive comparison between the source and the derived table.

The most secure and enterprise-recommended authentication method for Databricks automation is OAuth token federation with service principals .

This configuration allows service principals (non-human identities) to authenticate using temporary OAuth access tokens from a trusted identity provider (such as Azure AD or AWS IAM federation). These tokens are short-lived and scoped, significantly reducing credential exposure risks.

By contrast, static client secrets (B) or PATs (C) are long-lived and require periodic manual rotation, increasing security vulnerability. Shared user accounts (D) violate least-privilege and auditability principles. Therefore, A provides the strongest, most compliant authentication model for automated CLI and CI/CD workflows.

The best practice in such scenarios is to ensure that production data is handled securely and with proper access controls. By granting only read access to production data in development and testing environments, it mitigates the risk of unintended data modification. Additionally, maintaining isolated databases for different environments helps to avoid accidental impacts on production data and systems.

In Databricks Lakehouse, to retain manually deleted or updated records in the raw_iot table while recomputing downstream tables when a pipeline update is run, the property pipelines.reset.allowed should be set to false . This property prevents the system from resetting the state of the table, which includes the removal of the history of changes, during a pipeline update. By keeping this property as false, any changes to the raw_iot table, including manual deletes or updates, are retained, and recomputation of downstream tables, such as bpm_stats , can occur with the full history of data changes intact.

This is the correct answer because it exemplifies best practices for implementing this system. By isolating tables in separate databases based on data quality tiers, such as bronze, silver, and gold, the data engineering team can achieve several benefits. First, they can easily manage permissions for different users and groups through database ACLs, which allow granting or revoking access to databases, tables, or views. Second, they can physically separate the default storage locations for managed tables in each database, which can improve performance and reduce costs. Third, they can provide a clear and consistent naming convention for the tables in each database, which can improve discoverability and usability. Verified References: [Databricks Certified Data Engineer Professional], under “Lakehouse” section; Databricks Documentation, under “Database object privileges” section.

The error being raised is an AnalysisException, which is a type of exception that occurs when Spark SQL cannot analyze or execute a query due to some logical or semantic error 1 . In this case, the error message indicates that the query cannot resolve the column name ‘heartrateheartrateheartrate’ given the input columns ‘heartrate’ and ‘age’. This means that there is no column in the table named ‘heartrateheartrateheartrate’, and the query is invalid. A possible cause of this error is a typo or a copy-paste mistake in the query. To fix this error, the query should use a valid column name that exists in the table, such as ‘heartrate’. References:  AnalysisException

The requirement is that both valid (good) and invalid (bad) records must be captured and available separately for downstream processing. Invalid records should not simply be dropped; they must be quarantined in a dedicated table.

In Databricks Lakeflow Declarative Pipelines (DLT) , this is achieved by creating separate output tables :

One table for valid records (Silver table) that pass the expectations.

Another quarantine table that explicitly captures records failing the expectations.

Option A correctly implements this by:

Declaring a DLT table trips_data_quarantine.

Using .filter(expr(quarantine_rules)) to isolate invalid records (records where latitude or longitude is NULL).

This ensures analysts can query both good records (from the main Silver pipeline table) and bad records (from the quarantine table).

Why not the others?

B : Uses @dlt.expect_or_drop, which drops invalid records instead of quarantining them. This violates the requirement that quarantined data should be available.

C : Same as B, but applies expectations in bulk with expect_all_or_drop. Again, bad data is dropped, not quarantined.

D : Adds an is_quarantined flag in the same table. While it marks bad records, it does not separate them into a distinct quarantine table as required by the business use case.

Therefore, Option A is the only solution aligned with Databricks documentation for quarantining invalid data into a dedicated table while keeping valid data in the main pipeline.

 This is the correct answer because it addresses the situation while minimally interrupting other teams in the organization without increasing the number of tables that need to be managed. The situation is that an aggregate table used by numerous teams across the organization will need to have a number of fields renamed, and additional fields will also be added, due to new requirements from a customer-facing application. By configuring a new table with all the requisite fields and new names and using this as the source for the customer-facing application, the data engineering team can meet the new requirements without affecting other teams that rely on the existing table schema and name. By creating a view that maintains the original data schema and table name by aliasing select fields from the new table, the data engineering team can also avoid duplicating data or creating additional tables that need to be managed. Verified References: [Databricks Certified Data Engineer Professional], under “Lakehouse” section; Databricks Documentation, under “CREATE VIEW” section.