Design a Slack-like real-time messaging system
Company: OpenAI
Role: Software Engineer
Category: System Design
Difficulty: hard
Interview Round: Technical Screen
## System Design: Slack-like Messaging (DMs + Channels)
Design a Slack-like real-time messaging system that supports:
### Core features
- **Direct Messages (DMs)** between two users.
- **Channels** with many members (from small teams to **very large channels**).
- **Real-time delivery** to online users (web/mobile).
- **Multi-device**: the same user may be logged in on multiple devices simultaneously.
### Key design topics to cover
- How messages are **persisted** and **delivered** in real time.
- Push vs pull delivery models, including how you would handle **large channels**.
- Multi-device delivery semantics (e.g., once per device vs once per user) and how you track delivery/reads.
- Scalability: caching, partitioning/sharding, and how the system behaves under high fanout and spikes.
### Assumptions / constraints (you may refine)
- Message ordering should be correct **within a DM or channel**.
- The system should tolerate server failures and client reconnects.
- Latency target: near real-time for online users (e.g., p99 < 1–2s).
Provide APIs, data model, high-level architecture, and the key trade-offs/edge cases.
Quick Answer: This question evaluates a candidate's competency in designing scalable, fault-tolerant real-time messaging systems, including message persistence and delivery semantics for DMs and large channels, multi-device synchronization, ordering guarantees, and operational concerns like partitioning and latency targets.