System requirements

Functional:

1. 1-on-1 live chat
2. User can send text, image, video in the chat
3. Group chat should be able to send.

Non-Functional:

1. Low latency
2. High availability
3. Scalability
4. Strong consistency

Capacity estimation

Assume 1B daily active user

Read QPS: 1B / 100k = 10k

Peak read QPS: 2 * 10k = 20k

Assume 1% user send 10 message per day

Write QPS: 1B * 1% * 10 / 10k = 1k

Peak write QPS: 2 * 1k = 2k

Storage estimation:

The messages are stored client side. We only need to store user information, user friends information and group chat information. Also need to store unsent messages temporary.

Assume each user need to store 100kb information for un-sent user info, text message, images and videos temporary.

The storage estimation:

1B * 100kb = 100TB.

Each data will need 2 replicas. So in total the storage need is

100TB * 3 = 300TB.

API design

GET getAllMessages(userId)

WebSocket message:

From client to server:

sendMessage(messageType, fromId, toId, messageContent)

receiveMessage(messageType, fromId, toId, messageContent)

Database design

High-level design

Request flows

Detailed component design

Trade offs/Tech choices

Failure scenarios/bottlenecks

Future improvements