Optimize and Troubleshoot Data Storage and Data Processing DP-203 Practice Questions (Page 5)

Questions 21–25

1. 21expert · hard

   A Spark job in Azure Synapse Analytics fails with 'java.lang.OutOfMemoryError: Java heap space' during a join. The cluster has 10 executors with 8 GB memory each. The job reads two large tables and performs a join on a key that is not skewed. The data size is 100 GB per table. The job has already been optimized with appropriate shuffle partitions. Which additional action should the data engineer take?

   • AIncrease the executor memory to 16 GB.
   • BUse a broadcast join by increasing spark.sql.autoBroadcastJoinThreshold.
   • CIncrease spark.executor.memoryOverhead to 2 GB.
   • DIncrease spark.sql.shuffle.partitions further.
   Reveal Solution Discussion Select an answer first
2. 22application · medium

   A company runs a mix of interactive and batch workloads on an Azure Synapse Analytics dedicated SQL pool. The interactive queries are slow during batch loads. The team wants to ensure interactive queries get priority without changing the batch job's logic. Which feature should they configure?

   • AWorkload classification to assign interactive queries to a higher importance workload group.
   • BA non-clustered index on the tables used by interactive queries.
   • CResult-set caching for all queries.
   • DIncrease the DWU of the pool.
   Reveal Solution Discussion Select an answer first
4. 23expert · hard

   An Azure Data Factory pipeline fails at a Web activity that calls an external REST API. The error is 'Timeout'. The API is known to occasionally take over 2 minutes to respond. The pipeline is retried 3 times, but the activity still fails. The team wants to handle this without changing the API. Which configuration should be applied?

   • AAdd a retry policy with a longer interval.
   • BUse an Azure Function to call the API asynchronously.
   • CIncrease the Web activity's timeout setting to 5 minutes.
   • DChange the Web activity to a Lookup activity.
   Reveal Solution Discussion Select an answer first
5. 24application · medium

   A Spark job that aggregates sales data by product category is running slowly. The job processes a large dataset where a single category 'Electronics' accounts for 80% of the records. The cluster has 20 executors, but most executors finish quickly while one executor takes much longer. Which strategy should the data engineer use to balance the workload?

   • AEnable speculative execution on the cluster.
   • BIncrease the number of partitions to 200.
   • CAdd a salt key to the category column and perform two-level aggregation.
   • DUse a broadcast join for the category dimension table.
   Reveal Solution Discussion Select an answer first
6. 25foundation · easy

   In a Spark job, what does 'data spill' refer to?

   • AMoving data from a hot storage tier to a cold storage tier.
   • BDeleting data that is no longer needed from the data lake.
   • CWriting temporary data to disk when it does not fit in executor memory.
   • DReplicating data across multiple nodes for fault tolerance.
   Reveal Solution Discussion Select an answer first