Row key design strategies in Hadoop - Practice Problems & Coding Challenges

Challenge - 5 Problems

🧠 Conceptual

intermediate

Why is row key design important in Hadoop HBase?

In HBase, the design of the row key affects performance and data retrieval. Why is it important to choose a good row key?

A Because row keys control the replication factor of data in HDFS.

B Because row keys determine the file format used to store data on disk.

C Because a well-designed row key ensures even data distribution and fast lookups.

D Because row keys decide the compression algorithm applied to data.

💡 Hint

Think about how data is stored and accessed in HBase tables.

Row keys in HBase determine how data is sorted and distributed across regions. A good row key design avoids hotspots and improves query speed.

❓ Predict Output

intermediate

What is the output of this row key generation code?

Consider this Python code generating row keys for time-series data:

keys = [f"user_{i}_{20240101 + i}" for i in range(3)]
print(keys)

What is the output?

Hadoop

keys = [f"user_{i}_{20240101 + i}" for i in range(3)]
print(keys)

A ["user_0_20240101", "user_1_20240102", "user_2_20240103"]

B ["user_0_20240101", "user_1_20240101", "user_2_20240101"]

C ["user_0_20240101", "user_1_20240103", "user_2_20240105"]

D ["user_0_20240100", "user_1_20240101", "user_2_20240102"]

💡 Hint

Look at how the date part increments with i.

The date part adds i to 20240101, so it increases by 1 each time.

❓ data_output

advanced

How many regions will be created with this row key pattern?

In HBase, row keys are prefixed with a hash of the user ID modulo 4 to distribute load:

def generate_key(user_id, timestamp):
    prefix = user_id % 4
    return f"{prefix}_{user_id}_{timestamp}"

keys = [generate_key(i, 20240101) for i in range(8)]
regions = set(k.split('_')[0] for k in keys)
print(len(regions))

How many unique region prefixes are there?

Hadoop

def generate_key(user_id, timestamp):
    prefix = user_id % 4
    return f"{prefix}_{user_id}_{timestamp}"

keys = [generate_key(i, 20240101) for i in range(8)]
regions = set(k.split('_')[0] for k in keys)
print(len(regions))

A 1

B 8

C 2

D 4

💡 Hint

Consider the modulo operation and how many distinct values it produces for 0 to 7.

Modulo 4 of numbers 0 to 7 produces 4 distinct prefixes: 0,1,2,3.

🔧 Debug

advanced

Identify the error in this row key design code snippet

This code tries to create a row key by concatenating user ID and timestamp, but it raises an error:

user_id = 123
timestamp = 20240101
row_key = user_id + "_" + timestamp
print(row_key)

What error does this code raise?

Hadoop

user_id = 123
timestamp = 20240101
row_key = user_id + "_" + timestamp
print(row_key)

A NameError: name 'timestamp' is not defined

B TypeError: unsupported operand type(s) for +: 'int' and 'str'

C SyntaxError: invalid syntax

D ValueError: invalid literal for int() with base 10

💡 Hint

Check the data types of variables used with the plus operator.

Adding an integer and a string directly causes a TypeError in Python.

🚀 Application

expert

Which row key design best avoids hotspotting for sequential timestamps?

You have a dataset with many records per user, each with a sequential timestamp. You want to design row keys in HBase to avoid hotspotting (too much load on one region). Which row key design is best?

A Use row keys as <code>hash(userID)_timestamp</code> where hash distributes users evenly.

B Use row keys as <code>reverse_timestamp_userID</code> where timestamp digits are reversed.

C Use row keys as <code>userID_timestamp</code> with timestamps increasing sequentially.

D Use row keys as <code>timestamp_userID</code> with timestamps increasing sequentially.

💡 Hint

Think about how to spread writes evenly across regions.

Hashing userID prefixes the row key with a value that distributes data across regions, avoiding hotspotting caused by sequential timestamps.