Practice/Oracle/Design Meta Live Comments

Design Meta Live Comments

System DesignMust

Problem Statement

Design a real-time commenting system that allows users to view and post comments instantly while watching live videos or browsing posts. The system should handle massive-scale traffic and support features like reactions, live notifications, and seamless user transitions between different content streams.

The core challenge is designing low-latency fanout at massive scale, reasoning about hot-spot contention on a single stream, and handling unreliable mobile clients that disconnect, background, and switch streams rapidly. Strong answers connect product needs (smooth scrolling, no gaps or duplicates, graceful degradation) to the right architectural patterns: pub/sub, cursors, replay buffers, backpressure, and client-side strategies.

Key Requirements

Functional

Post comments -- users can post comments and reactions to an ongoing live video or content stream in real time
Real-time delivery -- new comments appear on all connected clients without manual refresh, streaming in sub-second
Catch-up on join -- users joining a stream see recent comments from before they connected to get context
Stream switching -- users can switch between live videos and immediately see only the relevant comments without gaps or duplicates

Non-Functional

Scalability -- support millions of concurrent viewers across thousands of simultaneous live streams with extreme skew on popular streams
Reliability -- handle mobile client disconnects and reconnects gracefully without losing comments or showing duplicates
Latency -- propagate new comments to all viewers within 500ms under normal conditions
Consistency -- eventual consistency is acceptable; all viewers should converge to the same comment timeline

What Interviewers Focus On

Based on real interview experiences, these are the areas interviewers probe most deeply:

1. Real-Time Fan-Out at Scale

Interviewers want to see how you'll deliver comments to millions of viewers on a single hot stream without overwhelming servers or creating bottlenecks.

Hints to consider:

Use a pub/sub system keyed by stream ID to route comments only to interested subscribers
Co-locate viewers of the same stream on the same WebSocket servers to minimize cross-server fan-out
Implement tiered fan-out: comments go to a message broker, then to edge servers, then to connected clients
Apply backpressure and sampling for extremely hot streams where comment volume exceeds what users can read

2. Client Reconnection and Catch-Up

Mobile clients frequently disconnect due to network changes, app backgrounding, or stream switching. Interviewers probe how you maintain a coherent experience across these disruptions.

Hints to consider:

Assign sequence numbers to each comment so clients can detect gaps and request retransmissions
Maintain a bounded replay buffer per stream (e.g., last 5 minutes) for fast catch-up on reconnect
Send the client's last-seen sequence number on reconnect to stream only missing comments
Handle deduplication on the client side using comment IDs to prevent duplicates during reconnection

3. Hot-Stream Contention

A single celebrity live stream can attract millions of viewers and thousands of comments per second, creating extreme contention on a single key.

Hints to consider:

Shard the fan-out for hot streams across multiple broker partitions or pub/sub channels
Use consistent hashing to distribute viewers across multiple WebSocket servers while keeping stream affinity
Implement comment rate limiting and moderation queues to manage comment volume on popular streams
Consider viewer-side sampling where not every comment is shown if the volume exceeds readable rates

4. Comment Persistence and Ordering

Interviewers look for how you balance real-time delivery speed with durable storage and consistent ordering.

Hints to consider:

Write comments to a durable append-only log (Kafka) for persistence and replay capability
Use server-assigned timestamps or sequence numbers for canonical ordering across distributed servers
Store comments in a database asynchronously from the real-time delivery path to avoid blocking
Design the read path to merge real-time streaming comments with persisted history for users who scroll back

Practice/Oracle/Design Meta Live Comments

Design Meta Live Comments

System DesignMust

Problem Statement

Key Requirements

Functional

Post comments -- users can post comments and reactions to an ongoing live video or content stream in real time
Real-time delivery -- new comments appear on all connected clients without manual refresh, streaming in sub-second
Catch-up on join -- users joining a stream see recent comments from before they connected to get context
Stream switching -- users can switch between live videos and immediately see only the relevant comments without gaps or duplicates

Non-Functional

Scalability -- support millions of concurrent viewers across thousands of simultaneous live streams with extreme skew on popular streams
Reliability -- handle mobile client disconnects and reconnects gracefully without losing comments or showing duplicates
Latency -- propagate new comments to all viewers within 500ms under normal conditions
Consistency -- eventual consistency is acceptable; all viewers should converge to the same comment timeline

What Interviewers Focus On

Based on real interview experiences, these are the areas interviewers probe most deeply:

1. Real-Time Fan-Out at Scale

Interviewers want to see how you'll deliver comments to millions of viewers on a single hot stream without overwhelming servers or creating bottlenecks.

Hints to consider:

Use a pub/sub system keyed by stream ID to route comments only to interested subscribers
Co-locate viewers of the same stream on the same WebSocket servers to minimize cross-server fan-out
Implement tiered fan-out: comments go to a message broker, then to edge servers, then to connected clients
Apply backpressure and sampling for extremely hot streams where comment volume exceeds what users can read

2. Client Reconnection and Catch-Up

Mobile clients frequently disconnect due to network changes, app backgrounding, or stream switching. Interviewers probe how you maintain a coherent experience across these disruptions.

Hints to consider:

Assign sequence numbers to each comment so clients can detect gaps and request retransmissions
Maintain a bounded replay buffer per stream (e.g., last 5 minutes) for fast catch-up on reconnect
Send the client's last-seen sequence number on reconnect to stream only missing comments
Handle deduplication on the client side using comment IDs to prevent duplicates during reconnection

3. Hot-Stream Contention

A single celebrity live stream can attract millions of viewers and thousands of comments per second, creating extreme contention on a single key.

Hints to consider:

Shard the fan-out for hot streams across multiple broker partitions or pub/sub channels
Use consistent hashing to distribute viewers across multiple WebSocket servers while keeping stream affinity
Implement comment rate limiting and moderation queues to manage comment volume on popular streams
Consider viewer-side sampling where not every comment is shown if the volume exceeds readable rates

4. Comment Persistence and Ordering

Interviewers look for how you balance real-time delivery speed with durable storage and consistent ordering.

Hints to consider:

Write comments to a durable append-only log (Kafka) for persistence and replay capability
Use server-assigned timestamps or sequence numbers for canonical ordering across distributed servers
Store comments in a database asynchronously from the real-time delivery path to avoid blocking
Design the read path to merge real-time streaming comments with persisted history for users who scroll back