Unit 1: Introduction to Spark - Practice Quiz

1 What is a common limitation of Hadoop MapReduce?

2 Why can repeated MapReduce jobs be slow?

3 Which type of workload is difficult for traditional MapReduce?

4 What does batch analytics do?

5 What is a key characteristic of real-time analytics?

6 Which example best represents batch analytics?

7 Which application commonly requires real-time analytics?

8 What is stream processing?

9 Which is an example of stream processing?

10 What does in-memory processing mean?

11 Why can in-memory processing improve performance?

12 Apache Spark is primarily a platform for what purpose?

13 Which feature allows Spark to reuse data efficiently?

14 Which Spark component is used for SQL queries?

15 Which Spark component supports machine learning tasks?

16 Which software is generally required before installing Apache Spark?

17 Which command commonly starts the Spark interactive shell for Scala?

18 What is the purpose of setting the SPARK_HOME environment variable?

19 Which Hadoop component is commonly used for distributed file storage?

20 Compared with traditional MapReduce, Spark is often faster because it can do what?

21 A machine learning algorithm repeatedly processes the same dataset for 50 iterations. Why is traditional Hadoop MapReduce likely to perform poorly for this workload?

22 A data-processing workflow contains ten dependent MapReduce jobs. Which limitation is most likely to increase the total execution time?

23 Which workload best illustrates a weakness of the MapReduce programming model?

24 A bank wants to block suspicious card transactions within two seconds of their occurrence. Which processing approach is most appropriate?

25 A retailer calculates monthly revenue from completed sales records. Which characteristic supports using batch analytics?

26 Which statement best compares batch and real-time analytics for a temperature-monitoring system?

27 An application receives click events continuously and updates a dashboard every five seconds. Which technique is being applied?

28 Why can caching a repeatedly used dataset in Spark memory improve an iterative analysis?

29 A streaming application must calculate the number of unique users observed during each ten-minute interval. Which concept is most directly required?

30 Which Spark feature allows the same application to combine SQL queries, machine learning, and streaming operations?

31 A Spark transformation is defined but no output is requested. Why might no computation occur immediately?

32 A Spark job fails after a worker loses a partition that was derived from source data. Which feature can help Spark recover the missing partition?

33 Which situation most strongly favors Spark over a disk-oriented batch engine?

34 A user wants to run Spark locally without installing Hadoop. Which deployment mode is most appropriate for initial learning and testing?

35 After installing Spark, a user runs a Spark application and receives an error indicating that Java cannot be found. Which configuration should be checked first?

36 Which command is commonly used to verify that a local Spark installation can start an interactive Python shell?

37 A standalone user wants to execute a Python Spark program from a terminal. Which tool is designed for submitting the application?

38 Which comparison between Spark and Hadoop MapReduce is most accurate?

39 A company already uses HDFS for storage but wants faster iterative analytics. Which adoption strategy is reasonable?

40 Which workload is generally a better fit for Spark than traditional Hadoop MapReduce?

41 A machine-learning algorithm performs 30 iterations over the same training data. Each iteration reads the previous iteration's output and produces a new model. Why is classic Hadoop MapReduce generally slower than Spark for this workload?

42 A workflow contains filter → map → filter → reduceByKey. In classic MapReduce, implementing each transformation as a separate job produces unnecessary overhead. Which Spark capability most directly reduces this overhead?

43 A MapReduce application receives continuously arriving events but launches one job every minute. Processing takes 40 seconds and job startup takes 20 seconds. What is the most fundamental obstacle to sub-second result latency?

44 A fraud system must block suspicious payments within 500 ms, while its detection model is retrained nightly from six months of transaction history. Which architecture best matches these requirements?

45 Events may arrive up to 10 minutes late. A dashboard computes five-minute event-time windows and must eventually include late events without retaining state forever. Which design is most appropriate?

46 A stream receives events at an average rate of 50,000 records/s. A micro-batch interval is 2 seconds, but each micro-batch requires 3 seconds to complete under sustained load. What is the most likely long-term behavior?

47 A stateful streaming job counts purchases per customer. After a failure, some input records may be replayed. Which combination provides the strongest basis for exactly-once observable results?

48 Which workload is most likely to gain the greatest benefit from explicitly caching a Spark dataset?

49 An RDD is reused by several actions, but it is larger than aggregate executor memory. Which persistence strategy best avoids recomputation while tolerating memory pressure?

50 A Spark program defines several transformations on a dataset, but no cluster work begins until count() is called. Which Spark design principle explains this behavior?

51 An executor loses three cached partitions of an RDD, but the source data remains available. Under normal conditions, how does Spark recover those partitions?

52 Which transformation most directly introduces a wide dependency and therefore normally creates a shuffle boundary between Spark stages?

53 A Spark job joins a 5 TB fact table with a 20 MB dimension table. Both are initially partitioned incompatibly. Which optimization most directly avoids shuffling the fact table?

54 Spark is unpacked in /opt/spark, but running spark-shell from another directory returns command not found. Java is already installed. Which configuration most directly fixes the issue for the current user?

55 A user runs spark-submit --master local[*] app.py on a machine that is also configured as a Spark standalone worker. Where will the application's tasks execute?

56 A standalone Spark master is reachable at spark://master.example:7077. A submission using --master local[8] never appears in the master's web interface. What change is required?

57 In Spark standalone mode, an application is submitted with deploy mode client. The user's terminal process is then terminated unexpectedly. Which outcome is most likely?

58 Which statement most accurately describes the relationship between Apache Spark and the Hadoop ecosystem?

59 An organization stores data in HDFS and manages resources with YARN. It wants to migrate computation from MapReduce to Spark while retaining storage and resource management. Which change is sufficient in principle?

60 A 100 TB archival transformation runs once per month, performs one sequential pass, and has no interactive or iterative reuse. Which conclusion about choosing Spark over MapReduce is most defensible?