Bird
Raised Fist0
LLDsystem_design~10 mins

Notification to all parties in LLD - Scalability & System Analysis

Choose your learning style10 modes available

Start learning this pattern below

Jump into concepts and practice - no test required

or
Recommended
Test this pattern10 questions across easy, medium, and hard to know if this pattern is strong
Scalability Analysis - Notification to all parties
Growth Table: Notification to All Parties
ScaleUsersNotifications/SecondSystem Changes
Small10010-50Single server handles notification dispatch; simple queue; direct DB writes
Medium10,0001,000-5,000Introduce message queue (e.g., Kafka); use worker pool; caching user preferences
Large1,000,000100,000+Multiple distributed queues; microservices for notification types; horizontal scaling; CDN for push notifications
Very Large100,000,00010,000,000+Global distributed system; sharded databases; multi-region queues; edge computing; advanced rate limiting
First Bottleneck

The first bottleneck is the message queue and notification dispatch system. As user count and notification volume grow, the queue can become overwhelmed, causing delays. Also, the database storing notification status can become a bottleneck due to high write/read load.

Scaling Solutions
  • Horizontal scaling: Add more worker servers to process notifications concurrently.
  • Message queue partitioning: Use multiple partitions or topics to distribute load.
  • Caching: Cache user notification preferences to reduce DB hits.
  • Sharding: Split notification data by user ID ranges or regions.
  • CDN and push services: Use content delivery networks and push notification services to offload delivery.
  • Rate limiting and batching: Control notification bursts and batch notifications to reduce load.
Back-of-Envelope Cost Analysis

At 1 million users sending 100,000 notifications per second:

  • Requests per second: ~100,000 (notification dispatch calls)
  • Storage: Assuming 1 KB per notification record, 100,000 notifications/sec = ~100 MB/sec = ~8.6 TB/day
  • Network bandwidth: 100,000 notifications * 1 KB = ~100 MB/s (800 Mbps), requires high bandwidth servers and network infrastructure
Interview Tip

Start by clarifying notification types and delivery guarantees. Discuss expected user scale and notification volume. Identify bottlenecks early (queue, DB, network). Propose incremental scaling steps: caching, horizontal scaling, sharding. Mention trade-offs like latency vs cost. Use real numbers to justify design choices.

Self Check

Your database handles 1000 QPS for notification status updates. Traffic grows 10x to 10,000 QPS. What do you do first?

Answer: Introduce read replicas and caching to reduce DB load, and consider sharding notification data to distribute writes. Also, optimize notification processing to reduce unnecessary DB writes.

Key Result
Notification systems first hit bottlenecks at message queues and databases as user and notification volume grow; scaling requires horizontal workers, queue partitioning, caching, and sharding.

Practice

(1/5)
1.

What is the main purpose of a notification system that sends messages to all parties?

easy
A. To quickly share important messages with everyone involved
B. To store large amounts of data securely
C. To perform complex calculations on user data
D. To create user profiles and preferences

Solution

  1. Step 1: Understand the role of notifications

    Notifications are designed to deliver messages to users or parties quickly and efficiently.
  2. Step 2: Identify the main goal

    The main goal is to share important information with all involved parties without delay.
  3. Final Answer:

    To quickly share important messages with everyone involved -> Option A
  4. Quick Check:

    Notification purpose = quick message sharing [OK]
Hint: Notifications = fast message delivery to all involved [OK]
Common Mistakes:
  • Confusing notifications with data storage
  • Thinking notifications perform data processing
  • Assuming notifications create user profiles
2.

Which of the following is the correct way to represent a notification service that sends messages to all parties in pseudocode?

function notifyAll(parties, message) {
  for (let i = 0; i < parties.length; i++) {
    parties[i].send(message);
  }
}
easy
A. Call send(message) once without looping
B. Send message only to the first party
C. Loop through parties and call send(message) on each
D. Loop through parties but do not send any message

Solution

  1. Step 1: Analyze the pseudocode loop

    The code loops through each party in the parties list using a for loop.
  2. Step 2: Check the send method call

    Inside the loop, it calls send(message) on each party, ensuring all get notified.
  3. Final Answer:

    Loop through parties and call send(message) on each -> Option C
  4. Quick Check:

    Loop + send call = notify all [OK]
Hint: Loop through all parties to send message [OK]
Common Mistakes:
  • Sending message only once
  • Not looping through all parties
  • Calling send outside the loop
3.

Consider this code snippet for notifying parties:

parties = ["Alice", "Bob", "Charlie"]
function notifyAll(parties, message) {
  let notified = []
  for (const person of parties) {
    notified.push(person + ": " + message)
  }
  return notified
}

console.log(notifyAll(parties, "Meeting at 5 PM"))

What will be the output?

medium
A. ["Alice: Meeting at 5 PM", "Bob: Meeting at 5 PM", "Charlie: Meeting at 5 PM"]
B. ["Meeting at 5 PM", "Meeting at 5 PM", "Meeting at 5 PM"]
C. ["Alice", "Bob", "Charlie"]
D. Error: notifyAll is not defined

Solution

  1. Step 1: Understand the loop behavior

    The function loops over each person in parties and creates a string combining their name and the message.
  2. Step 2: Check the returned list

    The notified list contains strings like "Alice: Meeting at 5 PM" for each party.
  3. Final Answer:

    ["Alice: Meeting at 5 PM", "Bob: Meeting at 5 PM", "Charlie: Meeting at 5 PM"] -> Option A
  4. Quick Check:

    Loop + string concat = list of personalized messages [OK]
Hint: Each party gets message with their name prefixed [OK]
Common Mistakes:
  • Returning only messages without names
  • Returning original party list
  • Assuming function is undefined
4.

Identify the bug in this notification function and select the fix:

function notifyAll(parties, message) {
  for (let i = 0; i < parties.length; i++) {
    parties.send(message)
  }
}
medium
A. Remove the loop and call parties.send(message) once
B. Add a return statement inside the loop
C. Change i < parties.length to i <= parties.length
D. Change parties.send(message) to parties[i].send(message)

Solution

  1. Step 1: Identify incorrect method call

    The code calls send on the entire parties array instead of individual party objects.
  2. Step 2: Fix by indexing the array

    Use parties[i].send(message) to call send on each party in the loop.
  3. Final Answer:

    Change parties.send(message) to parties[i].send(message) -> Option D
  4. Quick Check:

    Call send on each party object [OK]
Hint: Call send on parties[i], not parties array [OK]
Common Mistakes:
  • Calling send on the whole array
  • Using wrong loop condition
  • Adding unnecessary return inside loop
5.

You are designing a notification system to alert all parties involved in a project. Which design choice best ensures scalability and reliability?

  • A. Use a single server to send notifications sequentially to all parties.
  • B. Send notifications only to a random subset of parties to reduce load.
  • C. Store all notifications in a database and send them manually when needed.
  • D. Use a message queue to distribute notification tasks to multiple worker servers.
hard
A. Single server sending sequentially
B. Message queue with multiple workers
C. Store notifications and send manually
D. Send to random subset to reduce load

Solution

  1. Step 1: Evaluate single server approach

    Sending sequentially from one server limits scalability and can cause delays or failures.
  2. Step 2: Consider message queue with workers

    Using a message queue allows distributing notification tasks to multiple workers, improving scalability and reliability.
  3. Step 3: Assess other options

    Storing notifications for manual sending is slow; sending to random subset misses parties.
  4. Final Answer:

    Message queue with multiple workers -> Option B
  5. Quick Check:

    Queue + workers = scalable, reliable notifications [OK]
Hint: Use queues and workers for scalable notifications [OK]
Common Mistakes:
  • Relying on single server for all notifications
  • Sending notifications manually
  • Skipping parties to reduce load