Sample questions
Data Partitioning ScalingDifficulty 1
In data engineering, what does "partitioning" mean at a basic level?
- aEncrypting a table so only authorized users can read it
- bCompressing a table to reduce its disk footprint
- cSplitting data into smaller chunks based on a column's value✓
- dRenaming columns to follow a naming convention
Explanation:Partitioning divides a dataset into smaller chunks (partitions), typically based on a column's value range or category, so that data is physically or logically grouped.
Data Partitioning ScalingDifficulty 2
Why does partitioning typically help query performance?
- aIt automatically deletes old rows that are no longer needed
- bA query can skip scanning partitions it doesn't need✓
- cIt forces every query to scan the entire dataset for consistency
- dIt converts all columns to a single fixed data type
Explanation:When a query's filter aligns with the partition key, the engine can skip partitions that cannot contain matching rows, reducing the amount of data scanned and speeding up the query.
Data Partitioning ScalingDifficulty 2
What is the "partition key" in the context of partitioning?
- aThe primary key used to uniquely identify a row across the whole table
- bThe column whose value decides a row's partition✓
- cThe encryption key used to protect partition contents
- dThe name of the physical disk where the first partition is stored
Explanation:The partition key is the column (or combination of columns) whose values decide how rows are grouped into partitions, e.g. a date column or a region column.
Data Partitioning ScalingDifficulty 2
Which of the following is a reasonable basic criterion when choosing a partition key?
- aIt should be a column that is never used in any query filter
- bIt should always be a randomly generated value unrelated to the data
- cIt should be a column with only one possible value across the whole dataset
- dA column commonly used in filters that spreads data reasonably evenly✓
Explanation:A good partition key is frequently used in query filters (so pruning helps) and distributes rows across partitions without concentrating most rows into a single partition.
Data Partitioning ScalingDifficulty 1
What is "data skew" in the context of partitioning?
- aWhen data is stored in the wrong column type
- bWhen a table has no partition key defined at all
- cOne partition becomes much larger or busier than the others✓
- dWhen query results are returned in a different sort order than requested
Explanation:Data skew happens when rows are not evenly distributed across partitions, so one or a few partitions hold much more data or traffic than the rest.
Data Partitioning ScalingDifficulty 2
A dataset is partitioned by country. 90% of all rows belong to a single large country while the rest are spread across 40 smaller countries. What is this an example of?
- aPartition pruning working correctly
- bData skew, since one partition holds most of the data✓
- cA well-balanced partitioning scheme
- dA repartitioning operation in progress
Explanation:Because a single country partition holds the vast majority of rows, that partition is disproportionately large compared to the others — this is data skew.