Hot partition prevention in DynamoDB - Mini Project: Build & Apply

Hot Partition Prevention in DynamoDB

📖 Scenario: You are managing a DynamoDB table that stores user activity logs. The table uses UserID as the partition key. You notice that some users generate a lot of activity, causing hot partitions that slow down the database.To prevent this, you want to distribute the load more evenly by adding a suffix to the UserID to create multiple partitions per user.

🎯 Goal: Build a DynamoDB table design that prevents hot partitions by adding a partition key suffix. You will create the initial table schema, define a suffix count, implement logic to assign suffixes to UserID, and finalize the table with the new partition key.

📋 What You'll Learn

Create a DynamoDB table schema with UserID as the partition key and ActivityTimestamp as the sort key.

Define a variable suffix_count to control how many suffixes to use for partition keys.

Implement logic to assign a suffix to each UserID based on a hash modulo suffix_count.

Update the partition key to include the suffix, preventing hot partitions.

💡 Why This Matters

🌍 Real World

Hot partitions in DynamoDB can cause throttling and slow performance. This project shows how to design partition keys to distribute load evenly, which is critical for high-traffic applications like user activity tracking.

💼 Career

Database engineers and backend developers often need to optimize NoSQL database designs to handle uneven data access patterns. Understanding hot partition prevention is essential for building scalable, reliable systems.

1

Create DynamoDB table schema

Create a DynamoDB table schema named ActivityLog with UserID as the partition key and ActivityTimestamp as the sort key.

DynamoDB

# Define the DynamoDB table schema with UserID as partition key and ActivityTimestamp as sort key
# Your code here

Need a hint?

Use KeySchema to define UserID as the partition key (HASH) and ActivityTimestamp as the sort key (RANGE).

2

Define suffix count for partition keys

Define an integer variable called suffix_count and set it to 5. This will control how many suffixes to use for partition keys.

DynamoDB

ActivityLog = {
    'TableName': 'ActivityLog',
    'KeySchema': [
        {'AttributeName': 'UserID', 'KeyType': 'HASH'},
        {'AttributeName': 'ActivityTimestamp', 'KeyType': 'RANGE'}
    ],
    'AttributeDefinitions': [
        {'AttributeName': 'UserID', 'AttributeType': 'S'},
        {'AttributeName': 'ActivityTimestamp', 'AttributeType': 'S'}
    ],
    'ProvisionedThroughput': {
        'ReadCapacityUnits': 5,
        'WriteCapacityUnits': 5
    }
}

# Define suffix_count variable
# Your code here

Need a hint?

Set suffix_count to 5 to create 5 suffixes for partition keys.

3

Implement suffix assignment logic

Write a function called assign_partition_key that takes a user_id string and returns a new partition key by appending a suffix. The suffix is the hash of user_id modulo suffix_count. Use Python's built-in hash() function.

DynamoDB

ActivityLog = {
    'TableName': 'ActivityLog',
    'KeySchema': [
        {'AttributeName': 'UserID', 'KeyType': 'HASH'},
        {'AttributeName': 'ActivityTimestamp', 'KeyType': 'RANGE'}
    ],
    'AttributeDefinitions': [
        {'AttributeName': 'UserID', 'AttributeType': 'S'},
        {'AttributeName': 'ActivityTimestamp', 'AttributeType': 'S'}
    ],
    'ProvisionedThroughput': {
        'ReadCapacityUnits': 5,
        'WriteCapacityUnits': 5
    }
}

suffix_count = 5

def assign_partition_key(user_id):
    # Your code here

Need a hint?

Use hash(user_id) % suffix_count to get the suffix number, then append it to user_id with a # separator.

4

Update table schema with new partition key

Update the ActivityLog table schema to use PartitionKey as the new partition key instead of UserID. The PartitionKey will store the value returned by assign_partition_key. Keep ActivityTimestamp as the sort key.

DynamoDB

ActivityLog = {
    # Your code here
}

suffix_count = 5

def assign_partition_key(user_id):
    suffix = hash(user_id) % suffix_count
    return f"{user_id}#{suffix}"

Need a hint?

Change the partition key attribute name from UserID to PartitionKey in both KeySchema and AttributeDefinitions.