Fan-out on write vs fan-out on read in HLD - Architecture Patterns Compared

System Overview - Fan-out on write vs fan-out on read

This system demonstrates two common strategies for distributing updates to multiple users: fan-out on write and fan-out on read. The goal is to deliver timely updates to users' feeds while balancing system load and latency.

Fan-out on write pushes updates to all followers when a user posts, while fan-out on read fetches updates from all followed users when a user reads their feed.

Architecture Diagram

User
  |
  v
Load Balancer
  |
  v
API Gateway
  |
  +-----------------------+
  |                       |
Fan-out on Write Service  Fan-out on Read Service
  |                       |
  v                       v
Write DB               Read DB
  |                       |
  v                       v
Cache                   Cache
  |                       |
  +-----------+-----------+
              |
              v
           User Feed

Components

User

client

End user who reads and writes posts

Load Balancer

load_balancer

Distributes incoming user requests evenly to API Gateway

API Gateway

api_gateway

Routes requests to appropriate fan-out services

Fan-out on Write Service

service

Pushes new posts to all followers' feeds immediately on write

Fan-out on Read Service

service

Fetches posts from all followed users when a feed is requested

Write DB

database

Stores posts and follower relationships for fan-out on write

Read DB

database

Stores posts and follower relationships for fan-out on read

Cache

cache

Speeds up feed retrieval by caching user feeds or posts

User Feed

output

Final feed data delivered to the user

Request Flow - 10 Hops

User → Load Balancer

Load Balancer → API Gateway

API Gateway → Fan-out on Write Service

Fan-out on Write Service → Write DB

Fan-out on Write Service → Cache

API Gateway → Fan-out on Read Service

Fan-out on Read Service → Cache

Fan-out on Read Service → Read DB

Fan-out on Read Service → Cache

Fan-out on Read Service → User Feed

Failure Scenario

Component Fails:Cache

Impact:Feed reads become slower due to cache misses; writes still succeed but no fast feed updates

Mitigation:System falls back to database queries; cache rebuilds gradually; consider cache replication and monitoring

Architecture Quiz - 3 Questions

Which component immediately updates followers' feeds when a user posts?

A API Gateway

B Fan-out on Read Service

C Fan-out on Write Service

D Load Balancer

Design Principle

This architecture contrasts two fan-out strategies: pushing updates on write reduces read latency but increases write load, while pushing on read reduces write load but increases read latency. Caches help improve performance in both cases by reducing database hits.

Practice

1. What is the main advantage of using fan-out on write in system design?

easy

A. Simpler read logic by fetching data on demand

B. Faster read operations by duplicating data during write

C. Reduced storage usage by avoiding data duplication

D. Faster write operations by delaying data duplication

Solution

1. Step 1: Understand fan-out on write behavior

   Fan-out on write duplicates data to multiple places during the write operation.

2. Step 2: Analyze impact on read speed

   This duplication allows reads to be faster because data is already pre-distributed and ready to access.

3. Final Answer:

   Faster read operations by duplicating data during write -> Option B

4. Quick Check:

   Fan-out on write = Faster reads [OK]

Hint: Fan-out on write means write duplicates data for fast reads [OK]

Common Mistakes:

• Confusing fan-out on write with fan-out on read
• Thinking fan-out on write reduces storage
• Assuming writes are faster with fan-out on write

2. Which of the following best describes fan-out on read?

easy

A. Data is compressed during write to save storage

B. Data is duplicated during write to speed up reads

C. Data is cached permanently to reduce read latency

D. Data is fetched and combined during read to keep writes fast

Solution

1. Step 1: Define fan-out on read

   Fan-out on read means data is not duplicated during write but fetched from multiple sources during read.

2. Step 2: Understand write speed impact

   This keeps writes fast because no extra duplication work is done during write time.

3. Final Answer:

   Data is fetched and combined during read to keep writes fast -> Option D

4. Quick Check:

   Fan-out on read = Fast writes, complex reads [OK]

Hint: Fan-out on read delays data gathering until read time [OK]

Common Mistakes:

• Mixing fan-out on read with fan-out on write
• Assuming fan-out on read duplicates data during write
• Confusing caching with fan-out on read

3. Consider a system using fan-out on write. If a user updates their profile, what happens during the write operation?

medium

A. The update is duplicated to multiple storage locations immediately

B. The update is written once and read fetches combine data later

C. The update is cached temporarily and written later asynchronously

D. The update is compressed and stored in a single location

Solution

1. Step 1: Recall fan-out on write behavior

   Fan-out on write duplicates data during the write operation to multiple places.

2. Step 2: Apply to user profile update

   When a user updates their profile, the system writes the update to all relevant storage locations immediately.

3. Final Answer:

   The update is duplicated to multiple storage locations immediately -> Option A

4. Quick Check:

   Fan-out on write = Immediate duplication on write [OK]

Hint: Fan-out on write duplicates data immediately on update [OK]

Common Mistakes:

• Thinking update is written once and combined later
• Confusing caching with fan-out on write
• Assuming asynchronous write in fan-out on write

4. A system using fan-out on read is experiencing slow response times. What is a likely cause?

medium

A. Writes are slow due to data duplication

B. Storage is overloaded due to duplicated data

C. Reads are slow because data is fetched from multiple sources on demand

D. Data is compressed causing decompression delays

Solution

1. Step 1: Understand fan-out on read read behavior

   Fan-out on read fetches data from multiple sources during read, which can add latency.

2. Step 2: Analyze slow response cause

   Because reads combine data on demand, slow response times are likely due to this complex read process.

3. Final Answer:

   Reads are slow because data is fetched from multiple sources on demand -> Option C

4. Quick Check:

   Fan-out on read = Slow reads if sources are many [OK]

Hint: Fan-out on read can cause slow reads due to multiple fetches [OK]

Common Mistakes:

• Blaming slow writes in fan-out on read
• Assuming storage overload in fan-out on read
• Confusing compression delays with fan-out issues

5. You are designing a social media feed system. Which approach is better if you want instant feed updates but can tolerate higher storage costs?

hard

A. Fan-out on write to duplicate feed data for fast reads

B. Fan-out on read to keep writes fast and storage low

C. Use caching only without fan-out

D. Compress data on write to save storage

Solution

1. Step 1: Identify system needs

   Instant feed updates require fast reads with up-to-date data.

2. Step 2: Match approach to needs

   Fan-out on write duplicates feed data during write, enabling fast reads and instant updates but uses more storage.

3. Step 3: Evaluate other options

   Fan-out on read delays data gathering to read time, causing slower reads. Caching alone may not guarantee instant updates. Compression saves storage but slows access.

4. Final Answer:

   Fan-out on write to duplicate feed data for fast reads -> Option A

5. Quick Check:

   Instant updates + higher storage = Fan-out on write [OK]

Hint: Instant reads with more storage? Choose fan-out on write [OK]

Common Mistakes:

• Choosing fan-out on read for instant updates
• Ignoring storage cost impact
• Assuming caching replaces fan-out needs