Cache-aside pattern in HLD - System Design Guide

Problem Statement

When a system reads data directly from a slow database for every request, response times increase and the database can become overloaded under high traffic. This leads to poor user experience and potential downtime during traffic spikes.

Solution

The cache-aside pattern solves this by having the application check a fast cache first before querying the database. If the data is missing in the cache, the application loads it from the database and then stores it in the cache for future requests. This reduces database load and improves response times by serving frequent reads from the cache.

Architecture

Client

→Application

↓

Database

This diagram shows the client sending requests to the application, which first checks the cache. On a cache miss, the application queries the database and updates the cache before responding.

Trade-offs

✓ Pros

→

Reduces database load by serving frequent reads from cache.

→

Improves response times for cached data.

→

Simple to implement without requiring cache to be always consistent.

→

Allows selective caching of hot data.

✗ Cons

→

Cache can become stale if data changes frequently and cache invalidation is not handled.

→

Cache miss adds latency due to extra database query.

→

Requires application logic to manage cache reads and writes.

Use when read traffic is high and data can tolerate some staleness, typically over 1000 requests per second. Suitable when database is a bottleneck and caching hot data improves performance.

Avoid when data must always be strongly consistent or when write traffic is very high causing frequent cache invalidations that negate caching benefits.

Real World Examples

Amazon

Amazon uses cache-aside to reduce database load for product catalog reads, improving page load times during high traffic sales events.

Netflix

Netflix applies cache-aside to cache user session data and movie metadata, reducing latency and database stress during peak streaming hours.

Twitter

Twitter uses cache-aside to cache user timelines and tweets, enabling fast read access while keeping the database scalable.

Alternatives

Write-through cache

Cache is updated synchronously on every write, ensuring cache and database are always consistent.

Use when: Choose when strong consistency between cache and database is required and write latency overhead is acceptable.

Read-through cache

Cache automatically loads data from the database on a cache miss without application intervention.

Use when: Choose when you want to simplify application code by delegating cache loading to the caching system.

Write-back cache

Writes are made to cache first and asynchronously persisted to the database later.

Use when: Choose when write performance is critical and eventual consistency is acceptable.

Summary

Cache-aside pattern reduces database load by having the application load data into cache on demand.

It improves read performance by serving frequent requests from a fast cache while falling back to the database on misses.

This pattern requires application logic to manage cache reads, writes, and invalidations to avoid stale data.

Practice

1. What is the main purpose of the cache-aside pattern in system design?

easy

A. To load data into cache only when it is requested

B. To preload all data into cache before any request

C. To write data directly to cache without database interaction

D. To keep cache and database completely separate without synchronization

Solution

1. Step 1: Understand cache-aside pattern behavior

   The cache-aside pattern loads data into cache only when the application requests it and the data is not already cached.

2. Step 2: Compare options with this behavior

   To load data into cache only when it is requested correctly states this lazy loading approach. Other options describe preloading or unsynchronized caching, which are incorrect.

3. Final Answer:

   To load data into cache only when it is requested -> Option A

4. Quick Check:

   Cache-aside = lazy loading on demand [OK]

Hint: Cache-aside loads data only when requested [OK]

Common Mistakes:

• Thinking cache preloads all data
• Assuming cache updates happen before DB writes
• Believing cache and DB are unsynchronized

2. Which of the following is the correct sequence of operations in the cache-aside pattern when reading data?

easy

A. Read from database -> Update cache -> Return data

B. Read from cache -> If miss, read from database -> Update cache

C. Update cache -> Read from cache -> Read from database

D. Read from cache -> Return data without checking database

Solution

1. Step 1: Recall cache-aside read flow

   When reading, the system first checks the cache. If data is missing (cache miss), it reads from the database and then updates the cache.

2. Step 2: Match this flow to options

   Read from cache -> If miss, read from database -> Update cache matches this exact sequence. Other options either skip cache miss handling or update cache before reading.

3. Final Answer:

   Read from cache -> If miss, read from database -> Update cache -> Option B

4. Quick Check:

   Cache read -> DB read on miss -> Cache update [OK]

Hint: Cache read first, DB read only if cache miss [OK]

Common Mistakes:

• Updating cache before reading database
• Ignoring cache miss and reading only cache
• Reading database before checking cache

3. Consider this pseudocode using cache-aside pattern:

function getUser(id) {
  user = cache.get(id)
  if (user == null) {
    user = database.query(id)
    cache.set(id, user)
  }
  return user
}

What will getUser(5) return if the cache is empty and the database has user with id 5?

medium

A. null

B. Error due to cache miss

C. Empty object

D. User data from database

Solution

1. Step 1: Analyze cache miss handling

   Since cache is empty, cache.get(5) returns null, triggering database query.

2. Step 2: Database query and cache update

   The function queries the database for user id 5, finds the user data, updates the cache, and returns the user data.

3. Final Answer:

   User data from database -> Option D

4. Quick Check:

   Cache miss triggers DB read and cache update [OK]

Hint: Cache miss leads to DB read and cache update [OK]

Common Mistakes:

• Assuming cache miss returns null directly
• Expecting error on cache miss
• Ignoring cache update after DB read

4. A developer implemented cache-aside but notices stale data in cache after database updates. What is the likely mistake?

medium

A. Not updating cache after database write

B. Reading from cache before database

C. Preloading cache with all data

D. Using cache for write operations only

Solution

1. Step 1: Identify cause of stale cache

   Stale cache means cache has old data not reflecting recent database updates.

2. Step 2: Check cache update after DB write

   If cache is not updated or invalidated after database writes, stale data remains in cache.

3. Final Answer:

   Not updating cache after database write -> Option A

4. Quick Check:

   Cache must update after DB write to avoid stale data [OK]

Hint: Always update cache after DB write to avoid stale data [OK]

Common Mistakes:

• Ignoring cache invalidation after writes
• Assuming cache updates automatically
• Confusing read and write cache operations

5. You design a high-traffic product catalog service using cache-aside pattern. To handle frequent updates and reads, which strategy best maintains cache consistency and performance?

hard

A. Preload entire product catalog into cache at service start and never update cache

B. On product update, write to DB only and let cache expire naturally

C. On product update, write to DB then immediately update cache with new data

D. Write updates only to cache and asynchronously sync to DB later

Solution

1. Step 1: Understand cache consistency needs

   For frequent updates and reads, cache must reflect latest data quickly to avoid stale reads.

2. Step 2: Evaluate update strategies

   On product update, write to DB then immediately update cache with new data updates DB then cache immediately, ensuring cache consistency and fast reads. On product update, write to DB only and let cache expire naturally risks stale cache. Preload entire product catalog into cache at service start and never update cache ignores updates. Write updates only to cache and asynchronously sync to DB later risks data loss and inconsistency.

3. Final Answer:

   On product update, write to DB then immediately update cache with new data -> Option C

4. Quick Check:

   Update DB then cache for consistency and performance [OK]

Hint: Update DB then cache immediately for fresh data [OK]

Common Mistakes:

• Relying on cache expiration alone
• Preloading cache without updates
• Writing only to cache without DB sync