Partitioning & Sharding Fundamentals Quiz

Q1. Horizontal partitioning (dividing a table's rows across multiple servers) is commonly known as: Normalization
Sharding
Vertical partitioning
Replication

Q2. Which sharding strategy partitions data based on a user's geographic location (e.g., region)? Geo (zone) sharding
Hash-based sharding
Range-based sharding
Vertical partitioning

Q3. Which is a desirable property of an effective shard key? Values that increase in a predictable sequence
Only a few possible values for the key
Values closely correlated with insertion time
A high number of distinct key values

Q4. Which of the following is an example of vertical partitioning in a database? Partitioning a table's rows by user ID across multiple servers
Replicating the entire database to a second server
Splitting a table into multiple tables, each with a subset of the columns
Sharding a table by a range of values in its primary key

Q5. Range-based sharding is especially useful when: You need to efficiently retrieve a contiguous range of records by key
You have unpredictable access patterns and want even load distribution
You want to partition users by their geographic region
You need to split a table into multiple tables based on columns

Q6. Which of the following is a drawback of horizontal partitioning (sharding)? It prevents scaling the database beyond a single machine
It increases the number of columns in each table
Queries spanning multiple shards require complex coordination (scatter-gather)
It duplicates all data on every shard

Q7. What problem can occur if a timestamp or auto-incrementing ID is used as the shard key? Data will be duplicated on all shards
Queries by that key will no longer work
Data will distribute evenly across all shards
Most writes will target a single newest shard, causing a hotspot

Q8. What is a primary benefit of using consistent hashing for distributing data among shards? It guarantees perfectly uniform data distribution at all times
It minimizes how much data needs to be moved when adding or removing shards
It ensures every shard always stores an equal amount of data
It makes range queries on the shard key more efficient

Q9. After adding a new shard to a database cluster, how is rebalancing typically accomplished? Copy the full data set of an existing shard to the new shard
Switch to a consistent hashing scheme to avoid moving data
Migrate a subset of data (e.g., certain key ranges or chunks) from existing shards to the new shard
Do nothing; new data will only be written to the new shard

Q10. Which technique can help mitigate the issue of a hot partition key (a key receiving a disproportionate amount of traffic)? Adding a random prefix or salt to the key to spread the load
Using consistent hashing for all keys in the database
Partitioning the data vertically by splitting columns into different tables
Merging less active shards together into a single shard

Q11. Which metric would best indicate that one shard is handling a disproportionate load compared to others? The total QPS (queries per second) across the entire cluster
The average response time of the database as a whole
The number of records stored on each shard
The 99th percentile (p99) query latency on each shard

Q12. How can the progress of ongoing rebalancing (such as chunk migrations or partition splits) be monitored in a sharded system? By observing the average CPU utilization across the cluster
By tracking the number of remaining data chunks or split/merge tasks to be processed
By measuring the replication lag between shards
By checking the total data size of the largest shard

Q13. Which statement about Amazon DynamoDB partitioning is correct? DynamoDB automatically equalizes traffic across partitions, so no partition becomes hot
Adaptive capacity in DynamoDB is off by default and needs manual activation
Each DynamoDB partition can handle up to 3000 read capacity units (RCUs) and 1000 write capacity units (WCUs) per second, and adaptive capacity redistributes unused throughput to hot partitions
A single DynamoDB partition can grow to an unlimited size without splitting

Q14. Which is an advantage of using client-side sharding (with a shard map) over a single routing proxy? It removes the single point of failure and bottleneck that a central router introduces
Clients do not need any knowledge of the shard key or partition schema
Clients are guaranteed to always read the most up-to-date data
It greatly simplifies the client application logic

Q15. What does the key fan-out strategy for handling hot keys involve? Appending a random component to each key to randomize its distribution
Storing duplicates of a hot item under multiple different keys to spread load across shards
Caching the hot item in memory to reduce database load
Isolating the hot item on a single dedicated shard separate from other data

system-design