Data Operations and Support

146. An Amazon MWAA environment shows a DAG stuck in the queued state with tasks that never start. Which cause should the engineer investigate first?

1. A Check whether worker capacity and the DAG's concurrency settings allow additional tasks to be scheduled.
2. B Check whether the Amazon S3 bucket holding the DAG files has versioning enabled.
3. C Check whether the environment is running the latest available Apache Airflow version.
4. D Check whether the DAG file contains a Python syntax error that prevents it being parsed.

147. A data preparation step defined visually by analysts must run automatically as part of a nightly pipeline. Which solution meets these requirements?

1. A Build the step in Amazon QuickSight and export the prepared data each night.
2. B Build the step as an Amazon Athena view and query it from the pipeline each night.
3. C Build the step as an AWS Glue DataBrew recipe and run it as a scheduled DataBrew job within the pipeline.
4. D Have an engineer reimplement the analysts' logic as an AWS Glue Spark job.

148. An application must invoke AWS Glue and Amazon Athena operations programmatically from Python inside an existing service. Which solution meets these requirements?

1. A Schedule an AWS Glue job that performs the operations and have the application poll for completion.
2. B Shell out to the AWS CLI from the application and parse the command output.
3. C Create an Amazon API Gateway endpoint in front of each operation and call it over HTTP.
4. D Call the AWS SDK for Python from the application to invoke the Glue and Athena APIs.

149. A pipeline must run an AWS Glue job at 02:00 each day and also whenever a manifest file arrives in Amazon S3. Which solution meets these requirements?

1. A Create an AWS Glue scheduled trigger and have the pipeline poll the bucket for the manifest file.
2. B Create an Amazon EventBridge schedule for the daily run and an EventBridge rule on the S3 object created event, both targeting the job.
3. C Create two AWS Glue jobs, one scheduled and one triggered, that perform the same work.
4. D Create an Amazon MWAA DAG with a sensor task that waits for the manifest and a daily schedule.

150. An AWS Lambda function must transform each object as it lands in Amazon S3, and objects sometimes exceed the function's ephemeral storage. Which solution meets these requirements?

1. A Stream the object through the function in parts rather than downloading it whole, or mount Amazon EFS for the working files.
2. B Increase the function's memory allocation so more ephemeral storage is available.
3. C Reduce the S3 event notification batch size so fewer objects arrive at once.
4. D Increase the function's timeout so the object has more time to download.

151. A recurring Amazon Athena query must be run each morning and its results written to a known Amazon S3 location for downstream consumers. Which solution meets these requirements?

1. A Create an Athena view and have downstream consumers query it directly each morning.
2. B Ask an analyst to run the query in the Athena console each morning and copy the results.
3. C Create an AWS Glue crawler that runs each morning over the query's source tables.
4. D Schedule an AWS Lambda function with EventBridge that starts the query execution and writes results to the configured output location.

152. Business analysts must build interactive dashboards over data queried from Amazon Athena, with per-user access control. Which solution meets these requirements?

1. A Create an Amazon QuickSight dataset backed by Athena and control access through QuickSight users and groups.
2. B Create Amazon CloudWatch dashboards that display metrics from the Athena queries.
3. C Load the query results into Amazon OpenSearch Service and build dashboards there.
4. D Grant the analysts direct Athena query access and have them export results to spreadsheets.

153. Analysts repeatedly write the same complex three-table join and must be given a consistent way to query it in Amazon Athena. Which solution meets these requirements?

1. A Create an Athena CTAS query that materializes the join into a table each morning.
2. B Create an Athena view that encapsulates the join and grant the analysts access to the view.
3. C Ask each analyst to save the query in their own Athena query history.
4. D Create a separate Athena workgroup for each analyst with the query preloaded.

154. A data engineer must explore a large dataset interactively with Apache Spark without provisioning a cluster. Which solution meets these requirements?

1. A Run the exploration as an AWS Glue Python shell job writing results to Amazon S3.
2. B Launch an Amazon EMR cluster with Spark and attach a notebook to it.
3. C Run the exploration in an Amazon Athena notebook backed by Apache Spark.
4. D Run the exploration in an AWS Lambda function using Spark libraries.

155. A dataset contains duplicate rows and inconsistent capitalization in a category column, and analysts must clean it without writing code. Which solution meets these requirements?

1. A Build an AWS Glue DataBrew recipe that removes duplicates and standardizes the column, then run it as a job.
2. B Load the data into Amazon Redshift and clean it with a stored procedure.
3. C Write an AWS Glue Spark script that removes duplicates and standardizes the column.
4. D Create an Amazon Athena view that applies DISTINCT and a lower function to the column.

156. A report must present a seven-day rolling average of daily order volume from data held in Amazon Redshift. Which solution meets these requirements?

1. A Write a query using a window function with a seven-row preceding frame over the ordered daily totals.
2. B Export the daily totals to Amazon S3 and compute the average in an AWS Glue job.
3. C Write seven separate queries, one per day, and combine the results in the reporting tool.
4. D Write a query that groups by day and returns the totals for the reporting tool to average.

157. Analysts must be able to compare current month revenue against the same month last year, grouped by product line. Which solution meets these requirements?

1. A Write two separate queries and have the analysts compare the outputs manually.
2. B Write a query that aggregates by product line and period, using a self join or window function to align the two periods.
3. C Create a separate table for each month so the periods can be queried independently.
4. D Export both periods to Amazon S3 and compare them in a spreadsheet.

158. A team must be alerted when a nightly pipeline has not completed by 06:00. Which solution meets these requirements?

1. A Create a CloudWatch dashboard showing pipeline status and review it each morning.
2. B Publish a notification to Amazon SNS when the pipeline starts each night.
3. C Publish a completion metric from the pipeline and create a CloudWatch alarm that treats missing data as breaching within the window.
4. D Increase the pipeline's job timeout so it has more time to complete.

159. An Amazon Athena query that previously ran in seconds now takes minutes after the table accumulated millions of small files. Which solution meets these requirements?

1. A Add more columns to the table definition so the planner has better statistics.
2. B Compact the small files into objects of roughly 128 MB or larger and keep the partitions balanced.
3. C Increase the Athena query timeout so the query has time to complete.
4. D Change the query to use SELECT * so fewer planning decisions are required.

160. A team must be notified when an AWS Glue job fails or exceeds its expected duration, without polling the console. Which solution meets these requirements?

1. A Create an AWS Lambda function that polls the Glue API every minute for job status.
2. B Create an Amazon EventBridge rule on Glue job state change events that publishes to an Amazon SNS topic.
3. C Create an Amazon S3 event notification on the job's output prefix.
4. D Create a CloudWatch dashboard showing Glue job run history and review it each morning.

161. An engineer must diagnose a failed Apache Spark job on an Amazon EMR cluster that has already been terminated. Which combination of steps meets these requirements? (Select TWO.)

1. A Review the persistent application user interface or Spark History Server for stage and task detail.
2. B Review the AWS Config configuration history for the terminated cluster.
3. C Enable VPC Flow Logs on the cluster's subnet and analyse the network records.
4. D Review the cluster's billing detail for the period it ran.
5. E Configure a log URI on the cluster so application logs are archived to Amazon S3 as it runs.

162. A Kinesis consumer must be monitored to confirm it is keeping pace with the stream. Which metric should the team alarm on?

1. A Alarm on the number of shards configured on the stream.
2. B Alarm on IncomingBytes so producer volume is tracked.
3. C Alarm on the consumer instances' CPU utilization.
4. D Alarm on GetRecords.IteratorAgeMilliseconds so a rising lag is detected.

163. Pipeline costs have risen sharply and the team must attribute the increase to specific jobs. Which solution meets these requirements?

1. A Apply cost allocation tags to the pipeline resources, activate the tag keys, and analyse spend grouped by tag in AWS Cost Explorer.
2. B Delete the CloudWatch log groups belonging to the pipeline.
3. C Reduce every job's worker count by half and observe whether costs fall.
4. D Move all pipeline data to Amazon S3 Glacier Deep Archive.

164. An Amazon Redshift cluster runs both short interactive queries and long ETL jobs, and the interactive queries are being starved. Which solution meets these requirements?

1. A Configure workload management queues with separate concurrency and memory allocation for each workload.
2. B Apply a shorter statement timeout to every query on the cluster.
3. C Increase the cluster's node type so all queries complete more quickly.
4. D Create a materialized view over the ETL output so fewer queries run.

165. Logs from AWS Glue, Amazon EMR, and AWS Lambda must be searchable together during incident investigation. Which solution meets these requirements?

1. A Store each service's logs in a separate Amazon S3 bucket without an index.
2. B Review each service's console page separately during the incident.
3. C Send the logs to Amazon CloudWatch Logs and query across the log groups with CloudWatch Logs Insights.
4. D Download each service's logs and search them on a workstation.

166. An auditor must be able to establish which principal started a particular AWS Glue job run. Which solution meets these requirements?

1. A Query AWS CloudTrail for the StartJobRun event and read the calling identity from the record.
2. B Query Amazon CloudWatch Logs for the job's application output during the run.
3. C Review the AWS Config configuration timeline for the Glue job resource.
4. D Review the AWS Glue console job run history for the run identifier.

167. A batch must be rejected when more than 2 percent of rows have a null in a required column, and the pipeline must stop. Which solution meets these requirements?

1. A Create an AWS Lambda function that logs the null count to CloudWatch Logs.
2. B Create a CloudWatch dashboard that displays the null count for each batch.
3. C Evaluate an AWS Glue Data Quality ruleset as a pipeline step and branch the workflow on the result.
4. D Schedule a weekly manual review of the loaded data for null values.

168. A pipeline must confirm that a numeric column's values fall within an expected range before the data is published. Which solution meets these requirements?

1. A Sample ten rows after each run and inspect the values manually.
2. B Publish the data and add a dashboard widget showing the column's distribution.
3. C Define a data quality rule asserting the column's minimum and maximum, and evaluate it before the publish step.
4. D Define the column as a constrained type in the target table and rely on load failure.

169. An engineer must understand the distribution, cardinality, and null rates of an unfamiliar dataset before designing a pipeline for it. Which solution meets these requirements?

1. A Run a Glue crawler over the dataset and review the inferred schema.
2. B Run an AWS Glue DataBrew profile job over the dataset and review the resulting statistics.
3. C Query the first thousand rows with Amazon Athena and read them.
4. D Load the dataset into Amazon Redshift and inspect the table definition.

170. A Glue job must be started whenever a Glue crawler finishes successfully. Which approach is appropriate?

1. A Create a Glue workflow with a trigger that starts the job on crawler success.
2. B Schedule the job at a fixed time after the crawler's expected completion.
3. C Have the crawler invoke the job through a Lambda function.
4. D Run the job continuously and have it check for crawler completion.

171. A Glue DataBrew recipe must be applied to a new dataset each day automatically. Which approach is appropriate?

1. A Create a DataBrew job for the recipe and schedule it, or trigger it from the data's arrival event.
2. B Convert the recipe to a Spark script and schedule that.
3. C Apply the recipe manually and save the output.
4. D Ask the analyst to run the recipe each morning.

172. An Amazon Redshift stored procedure must run nightly without an external scheduler. Which approach is appropriate?

1. A Create a Lambda function on a schedule that calls the procedure.
2. B Create a scheduled query in Redshift that invokes the procedure.
3. C Create a Step Functions workflow on a schedule.
4. D Run the procedure manually each night.

173. An Athena query over a large table is scanning far more data than the filtered result requires. Which cause should be investigated first?

1. A Investigate whether the results bucket is in the same Region.
2. B Investigate whether the query uses SELECT *.
3. C Investigate whether the table is partitioned on the filter column and whether the files are in a columnar format.
4. D Investigate whether the workgroup has a data scanned limit.

174. A QuickSight dashboard must refresh its data automatically from a Redshift source. Which approach is appropriate?

1. A Export the Redshift data to CSV each morning and upload it.
2. B Ask the analyst to refresh the dashboard manually.
3. C Rebuild the dashboard each day.
4. D Create a SPICE dataset with a scheduled refresh, or use direct query for live data.

175. A data engineer must run an ad hoc SQL query across data in both Amazon S3 and Amazon Redshift. Which approach is appropriate?

1. A Load the S3 data into Redshift first.
2. B Export the Redshift data to S3 and query with Athena.
3. C Query from Redshift using Spectrum to reach the S3 data alongside the local tables.
4. D Query each source separately and combine in a spreadsheet.

176. A pipeline's Glue jobs are completing successfully but taking progressively longer each week. Which cause should be investigated first?

1. A Investigate whether the job's IAM role has been modified.
2. B Investigate whether the output bucket's storage class has changed.
3. C Investigate whether the Glue version has changed.
4. D Investigate whether the input data volume is growing or whether small files are accumulating in the source.

177. A data pipeline's cost must be attributed to the business team that owns each dataset. Which approach is appropriate?

1. A Attribute all cost to the platform team.
2. B Tag the pipeline's resources with the owning team and activate the tag for cost allocation.
3. C Review the monthly bill and estimate each team's share.
4. D Charge each team an equal share.

178. A Redshift query is slow, and the execution plan shows a nested loop join. Which cause should be investigated first?

1. A Investigate whether the table needs a VACUUM.
2. B Investigate whether the join condition is missing or non-equality, which forces a nested loop.
3. C Investigate whether the WLM queue is saturated.
4. D Investigate whether the cluster needs more nodes.

179. A streaming pipeline's Lambda consumer is being invoked but the iterator age keeps rising. Which cause should be investigated first?

1. A Investigate whether the stream has too few consumers registered.
2. B Investigate whether the function's memory is too high.
3. C Investigate whether the function's processing time per batch exceeds the rate at which records arrive, so it cannot keep up.
4. D Investigate whether the stream's retention is too short.

180. A pipeline must alert when a Glue job runs longer than its expected duration. Which approach is appropriate?

1. A Set a CloudWatch alarm on the job's CPU utilization.
2. B Increase the job's worker count so it finishes faster.
3. C Set a job timeout and route the timeout event to a notification through EventBridge.
4. D Review the job run history each morning.

181. A data quality check must run on a table after each load and record its results over time. Which approach is appropriate?

1. A Run a one-time profile when the table is created.
2. B Rely on downstream consumers to report problems.
3. C Inspect a sample manually after each load.
4. D Define a Glue Data Quality ruleset and evaluate it as a step after each load, publishing results as metrics.

182. A pipeline must quarantine records that fail a quality rule rather than failing the whole batch. Which approach is appropriate?

1. A Route failing records to a separate location with the rule violation attached, and continue processing the remainder.
2. B Drop the failing records silently.
3. C Correct the failing records automatically with default values.
4. D Fail the batch on the first failing record.

183. A quality rule must confirm that a column's values fall within an expected range, and the range shifts seasonally. Which approach is appropriate?

1. A Widen the fixed range to cover all seasons.
2. B Use a rule that compares against a statistical baseline learned from history rather than a fixed range.
3. C Disable the rule during seasons when it fails.
4. D Set the range to the annual average.

184. A pipeline must run when a specific object key pattern appears in a bucket, ignoring other objects. Which configuration is appropriate?

1. A Poll the bucket and check each key.
2. B Trigger on every object and filter inside the pipeline.
3. C Create a separate bucket for the matching objects.
4. D Create an EventBridge rule with an event pattern matching the prefix and suffix so only matching objects trigger the pipeline.

185. A daily pipeline must not run when the previous day's run has not completed. Which approach is appropriate?

1. A Rely on the schedule interval being longer than the typical run.
2. B Reduce the pipeline's runtime so overlap is unlikely.
3. C Increase the pipeline's concurrency so both can run.
4. D Check for an in-progress execution at the start and exit without starting a second one.

186. An Athena query must write its results to a specific location with encryption applied. Which configuration is appropriate?

1. A Copy the results to the required location after each query.
2. B Configure the workgroup's result location and encryption settings so all queries in the workgroup inherit them.
3. C Encrypt the results manually after each query.
4. D Specify the location in each query's client configuration.

187. A recurring analytical query must run faster without changing the underlying data. Which approach is appropriate?

1. A Run the query more frequently.
2. B Reduce the number of columns displayed in the result.
3. C Increase the query engine's timeout.
4. D Materialize the query's result as a table refreshed on a schedule and query that instead.

188. A data engineer must join data held in Amazon Redshift with data in an Amazon RDS database without moving either. Which capability is appropriate?

1. A A Glue job that joins the two and writes the result.
2. B A Redshift federated query, which reads the RDS data in place as part of the Redshift query.
3. C A nightly export of both to Amazon S3.
4. D An UNLOAD from Redshift followed by a load into RDS.

189. A Glue job fails with an out-of-memory error on the driver rather than an executor. Which cause should be investigated first?

1. A Investigate whether the job has too few executors.
2. B Investigate whether the script collects a large result to the driver, since driver memory is not sized for full datasets.
3. C Investigate whether job bookmarks are enabled.
4. D Investigate whether the source data is compressed.

190. An Athena query fails with an error about too many open partitions. Which cause should be investigated first?

1. A Investigate whether the table's partitioning scheme creates far more partitions than the query needs, such as partitioning by a high-cardinality column.
2. B Investigate whether the results bucket is full.
3. C Investigate whether the query uses too many columns.
4. D Investigate whether the workgroup has a data scanned limit.

191. A Redshift cluster's queries have slowed, and the tables have received large volumes of updates and deletes. Which maintenance operation is appropriate?

1. A Increase the cluster's node count.
2. B Run VACUUM to reclaim space from deleted rows and restore sort order, then ANALYZE to refresh statistics.
3. C Rebuild the tables from scratch.
4. D Reduce the number of concurrent queries.

192. A pipeline's cost must be tracked per pipeline rather than per service. Which approach is appropriate?

1. A Tag each pipeline's resources with a pipeline identifier, activate the tag, and group spend by it.
2. B Review the bill grouped by service.
3. C Create a separate account per pipeline.
4. D Estimate each pipeline's share from its runtime.

193. A data quality rule must confirm that a foreign key value exists in the referenced dataset. Which rule type is appropriate?

1. A A completeness rule checking for nulls.
2. B A range rule checking minimum and maximum values.
3. C A referential integrity rule comparing the column's values against the referenced dataset.
4. D A uniqueness rule checking for duplicates.

194. A data quality score must be tracked over time so gradual degradation is visible. Which approach is appropriate?

1. A Review the ruleset's output after each run.
2. B Record the result in the pipeline's log.
3. C Publish the ruleset's result as a metric after each run and chart it over time.
4. D Alert only when the ruleset fails entirely.

195. A pipeline must be triggered by the completion of another pipeline in a different account. Which approach is appropriate?

1. A Run both pipelines on the same schedule with an offset.
2. B Share credentials so the first pipeline starts the second directly.
3. C Publish a completion event to an event bus in the consuming account and trigger the pipeline from a rule there.
4. D Poll the other account's pipeline status on a schedule.

196. A pipeline must run only after all files for a batch have arrived, and the file count varies. Which approach is appropriate?

1. A Run the pipeline hourly and process whatever has arrived.
2. B Wait a fixed period after the first file arrives.
3. C Have the producer write a manifest listing the batch's files and trigger on the manifest.
4. D Trigger on each file's arrival and count them in the pipeline.

197. An automation must start a Glue job only when the previous run has finished. Which approach is appropriate?

1. A Increase the job's worker count so it finishes sooner.
2. B Rely on the schedule interval exceeding the job's duration.
3. C Check the job's current run state before starting and exit if a run is in progress.
4. D Set the job's maximum concurrent runs to a high value.

198. A scheduled pipeline must skip execution on non-business days. Which approach is appropriate?

1. A Disable the schedule manually on non-business days.
2. B Create a separate schedule for each business day.
3. C Evaluate the date against a business calendar at the start of the workflow and exit cleanly on non-business days.
4. D Run every day and let the job produce empty output.

199. A pipeline must notify a downstream team only when its output differs materially from the previous run. Which approach is appropriate?

1. A Notify on every successful run.
2. B Notify on a weekly schedule.
3. C Notify only on failure.
4. D Compare the run's summary statistics against the previous run and notify only when the difference exceeds a threshold.

200. An orchestration must run a step only when a previous step produced output. Which approach is appropriate?

1. A Have the previous step return a result the workflow branches on, so the dependent step is skipped when there is no output.
2. B Check for output manually before starting the workflow.
3. C Run the dependent step unconditionally and exit early inside it.
4. D Run the dependent step on a delay.

201. A pipeline must reprocess a specific historical date range on demand without changing its scheduled behaviour. Which approach is appropriate?

1. A Create a separate pipeline for reprocessing.
2. B Parameterize the pipeline's date range with the schedule supplying the default, so a manual run can override it.
3. C Temporarily change the schedule to the historical date.
4. D Edit the job's code for the reprocessing run.

202. An automated pipeline must stop and alert rather than proceed when its input volume is far below normal. Which approach is appropriate?

1. A Wait for more input to arrive.
2. B Reduce the pipeline's schedule frequency.
3. C Process whatever input has arrived.
4. D Compare the input volume against the recent baseline and fail the run with an alert when it falls outside the expected range.

203. A pipeline's steps must run in different accounts with the orchestration defined in one place. Which approach is appropriate?

1. A Define a separate workflow in each account.
2. B Grant the orchestration account administrator access to each target.
3. C Replicate the data into the orchestration account.
4. D Define the workflow centrally and have each step assume a role in the target account to perform its work.

204. An automation must apply the same transformation to data landing in several buckets. Which approach is appropriate?

1. A Create a separate workflow per bucket.
2. B Route events from each bucket to one workflow that takes the source location as a parameter.
3. C Run the transformation on a schedule across all buckets.
4. D Consolidate the data into one bucket first.

205. A pipeline must record whether each run processed new data or found nothing to do. Which approach is appropriate?

1. A Succeed silently when there is nothing to process.
2. B Log the outcome without emitting a metric.
3. C Emit a distinct outcome from the run and record it as a metric so both states are visible over time.
4. D Fail the run when there is nothing to process.

206. An automation must prevent two pipeline runs from writing to the same output partition simultaneously. Which approach is appropriate?

1. A Write each run's output to a new partition.
2. B Reduce the pipeline's schedule frequency.
3. C Write to the partition and resolve conflicts afterwards.
4. D Acquire a lock on the partition before writing and release it on completion, so a second run waits or exits.

207. A pipeline must be redeployed to a new environment with its schedules disabled until it is validated. Which approach is appropriate?

1. A Deploy with schedules enabled and disable them immediately.
2. B Deploy without schedules and add them manually later.
3. C Parameterize the schedule's enabled state so the deployment creates it disabled and it is enabled after validation.
4. D Deploy to the new environment only after validation elsewhere.

208. An orchestration must limit how many pipeline runs execute concurrently across an account. Which approach is appropriate?

1. A Run all pipelines sequentially in one workflow.
2. B Configure a concurrency limit on the workflow or use a semaphore pattern so waiting runs queue rather than all executing.
3. C Reduce each pipeline's schedule frequency.
4. D Increase the account's service quotas.

209. A pipeline must pass a large intermediate result between steps without exceeding the workflow's payload limit. Which approach is appropriate?

1. A Compress the result and pass it inline.
2. B Write the result to Amazon S3 and pass its location between steps rather than the data itself.
3. C Split the result across several workflow variables.
4. D Increase the workflow's payload limit.

210. An automation must retry a failed pipeline run from the failed step rather than from the beginning. Which approach is appropriate?

1. A Rerun the entire pipeline from the start.
2. B Increase the retry count on the whole pipeline.
3. C Reduce the pipeline to fewer, larger steps.
4. D Structure the pipeline as discrete states with checkpointed outputs so execution can resume at the failure point.

211. A pipeline's cost has risen after a source system began sending significantly more data. Which response is appropriate?

1. A Reduce the pipeline's schedule frequency.
2. B Move the output to a cheaper storage class.
3. C Sample the input to reduce volume.
4. D Confirm the additional data is needed, and if so right-size the pipeline's resources to the new volume rather than leaving them over- or under-provisioned.

212. A Glue job's logs must be retained longer than the default for compliance. Which approach is appropriate?

1. A Set the retention period on the job's log groups, or deliver the logs to Amazon S3 with a lifecycle policy.
2. B Enable job bookmarks.
3. C Increase the job's timeout.
4. D Increase the job's worker count.

213. A pipeline's performance must be compared before and after a change. Which approach is appropriate?

1. A Compare the two most recent runs.
2. B Record duration and resource metrics for a representative set of runs before the change and compare the same metrics afterwards.
3. C Compare the pipeline's cost for the month.
4. D Compare the change's code diff.

214. An Athena workgroup must prevent a single query from scanning more data than expected. Which configuration is appropriate?

1. A Reduce the workgroup's concurrency.
2. B Set a per-workgroup monthly limit only.
3. C Increase the query timeout.
4. D Set a per-query data scanned limit on the workgroup so a runaway query is cancelled.

215. A Redshift cluster's disk usage is growing although the data volume is stable. Which cause should be investigated first?

1. A Investigate whether the tables are compressed.
2. B Investigate whether the cluster needs more nodes.
3. C Investigate whether deleted rows have not been reclaimed, since space from updates and deletes is not released until maintenance runs.
4. D Investigate whether backups are consuming cluster storage.

216. A pipeline must alert when its output row count deviates materially from the historical pattern. Which approach is appropriate?

1. A Publish the row count as a metric and alarm on an anomaly detection band rather than a fixed threshold.
2. B Alarm on a fixed threshold set at project start.
3. C Alarm when the row count is zero.
4. D Review the row counts weekly.

217. A data quality rule must be applied to a column only when another column has a particular value. Which approach is appropriate?

1. A Apply the rule to every row and accept the failures.
2. B Split the dataset into two and apply the rule to one.
3. C Express the rule with a condition so it evaluates only the applicable subset of rows.
4. D Remove the rule entirely.