Why Eviction policies (LRU, LFU, random) in Redis? - Purpose & Use Cases

The Big Idea

What if your system could decide by itself which data to forget to keep running fast?

The Scenario

Imagine you have a small shelf to keep your favorite books, but you keep buying new books every week. When the shelf is full, you have to decide which book to remove to make space for the new one.

The Problem

Manually choosing which book to remove is slow and confusing. You might forget which books you read recently or which ones you like the most. This can lead to removing important books by mistake or wasting time deciding.

The Solution

Eviction policies like LRU (Least Recently Used), LFU (Least Frequently Used), and random automatically decide which items to remove when space runs out. This keeps your shelf organized without you having to think about it.

Before vs After

✗ Before

if shelf_full:
    ask_user_which_book_to_remove()

✓ After

if shelf_full:
    remove_item_using_LRU_policy()

What It Enables

It enables automatic, smart management of limited space so your system stays fast and efficient without manual effort.

Real Life Example

In Redis, when memory is full, eviction policies decide which cached data to remove so your app keeps running smoothly without crashing or slowing down.

Key Takeaways

Manual removal of items is slow and error-prone.

Eviction policies automate smart removal decisions.

This keeps systems efficient and reliable under memory limits.

Practice

1. Which Redis eviction policy removes the least recently used keys first when memory is full?

easy

A. LRU (Least Recently Used)

B. LFU (Least Frequently Used)

C. Random

D. No eviction

Solution

1. Step 1: Understand eviction policies in Redis

   Redis uses eviction policies to free memory when it is full by removing keys based on certain rules.

2. Step 2: Identify the policy that removes least recently used keys

   LRU stands for Least Recently Used, which means it removes keys that have not been accessed for the longest time.

3. Final Answer:

   LRU (Least Recently Used) -> Option A

4. Quick Check:

   Eviction policy removing least recently used keys = LRU [OK]

Hint: LRU = removes oldest unused keys first [OK]

Common Mistakes:

• Confusing LRU with LFU
• Thinking random eviction is based on usage
• Assuming no eviction means keys are removed

2. Which of the following is the correct Redis configuration directive to set the eviction policy to LFU?

easy

A. maxmemory-policy allkeys-lfu

B. maxmemory-policy random-eviction

C. maxmemory-policy volatile-lru

D. maxmemory-policy noeviction

Solution

1. Step 1: Recall Redis maxmemory-policy options

   Redis uses the 'maxmemory-policy' directive to set eviction behavior, with options like 'allkeys-lfu' for LFU eviction.

2. Step 2: Identify the correct syntax for LFU eviction

   The correct directive for LFU eviction on all keys is 'maxmemory-policy allkeys-lfu'.

3. Final Answer:

   maxmemory-policy allkeys-lfu -> Option A

4. Quick Check:

   LFU eviction config = maxmemory-policy allkeys-lfu [OK]

Hint: LFU config uses 'allkeys-lfu' in maxmemory-policy [OK]

Common Mistakes:

• Using 'random-eviction' instead of LFU
• Confusing 'volatile-lru' with LFU
• Missing 'maxmemory-policy' keyword

3. Given Redis is configured with the maxmemory-policy allkeys-lru and memory is full, what happens when a new key is added?

medium

A. A random key is evicted to make space

B. The least frequently used key is evicted to make space

C. The least recently used key is evicted to make space

D. The new key is rejected and not added

Solution

1. Step 1: Understand the 'allkeys-lru' eviction policy

   This policy evicts the least recently used key among all keys when memory is full.

2. Step 2: Predict behavior when adding a new key

   When memory is full, Redis removes the least recently used key to free space for the new key.

3. Final Answer:

   The least recently used key is evicted to make space -> Option C

4. Quick Check:

   allkeys-lru evicts least recently used key [OK]

Hint: allkeys-lru evicts oldest used key on full memory [OK]

Common Mistakes:

• Confusing LRU with LFU eviction
• Thinking random eviction applies here
• Assuming new keys are rejected automatically

4. You configured Redis with maxmemory-policy volatile-lfu but notice keys without expiration are never evicted. What is the likely cause?

medium

A. LFU policy does not work with volatile keys

B. Only keys with expiration are considered for eviction in volatile-lfu

C. Redis randomly evicts keys regardless of policy

D. Keys without expiration are evicted first

Solution

1. Step 1: Understand volatile-lfu eviction policy

   Volatile policies only evict keys that have an expiration set (TTL).

2. Step 2: Analyze why keys without expiration remain

   Since keys without expiration are not eligible, they are never evicted under volatile-lfu.

3. Final Answer:

   Only keys with expiration are considered for eviction in volatile-lfu -> Option B

4. Quick Check:

   Volatile policies evict only expiring keys [OK]

Hint: Volatile policies evict only keys with expiration [OK]

Common Mistakes:

• Assuming LFU ignores expiration
• Thinking random eviction applies here
• Believing keys without expiration are evicted first

5. You want Redis to evict keys randomly but only among keys with expiration set. Which eviction policy should you configure?

hard

A. volatile-lru

B. allkeys-random

C. allkeys-lfu

D. volatile-random

Solution

1. Step 1: Understand eviction policy prefixes

   'allkeys-' applies to all keys; 'volatile-' applies only to keys with expiration.

2. Step 2: Identify random eviction among volatile keys

   'volatile-random' evicts keys randomly but only among keys with expiration set.

3. Final Answer:

   volatile-random -> Option D

4. Quick Check:

   Random eviction on expiring keys = volatile-random [OK]

Hint: Volatile-random evicts random expiring keys only [OK]

Common Mistakes:

• Choosing allkeys-random which evicts all keys
• Confusing LFU or LRU with random eviction
• Selecting volatile-lru which is not random