System design interview question

Design a chat app

WhatsApp, Messenger, Slack: “design a chat app” tests whether you can keep millions of people connected at once, deliver a message in a fraction of a second, and still get it to someone whose phone is switched off. It is the classic question about real-time delivery. Here is how to walk into it, and then how to practise it until the answer is yours.

Practise Chat App

Multiple choicePick the design from four
Fill the gapsFill the gaps in one drawn for you
Build from scratchBuild it on a blank canvas

Practise it on a canvas

Free with an account.

The question, as an interviewer would set it

A messaging app lets people chat one-to-one and in small groups from their phones. A message sent to someone who has the app open should arrive within a fraction of a second, without their phone repeatedly asking the server whether anything is new. Hundreds of thousands of people are connected at any moment, spread across many servers. When the recipient's app is closed, the message must still reach them: their phone should get an alert, and the message must be waiting when they open the app, even days later. Messages arrive in enormous volume, are almost never edited, and are always read one conversation at a time, newest first.

What it is really testing

Every phrase that matters in the brief is a requirement in disguise. Spot them before you draw a single box.

“without their phone repeatedly asking the server whether anything is new”
A connection held open so the server can hand messages over as they arrive.
“spread across many servers”
A way for the server that receives a message to reach the one holding the recipient's connection.
“their phone should get an alert”
The phone platform's push service, the only way to reach a closed app.
“the message must be waiting when they open the app, even days later”
Durable storage of every message.
“always read one conversation at a time, newest first”

Back of the envelope

Daily active users: 500 millionAn assumption to say out loud. If the interviewer has a number in mind, they will give it to you.
Messages: ~40 a day each → ~20 billion a day, ~230,000 a secondThat is the average; evenings and big events run several times higher.
Connected at once: ~10% of daily users → ~50 million open connectionsEach is held for as long as the app is open, so this is a question of memory per machine, not requests per second.
New storage: ~100 bytes of text each → ~2 TB a dayBefore metadata and copies for safety; over the years it runs to petabytes.

Questions to ask first

One-to-one only, or groups too, and how big can a group get?
Must a message survive the recipient being offline for days, or can it expire?
Does the order of messages within a conversation have to be exact?
Do we need read receipts and online status, or only delivery?
Text only, or photos, voice notes and video as well?

The trade-offs you will be pushed on

1.How does the server tell a phone about a new message without the phone asking again and again?
2.The sender and the recipient are connected to different servers. How does a message get from one to the other?
3.What guarantees the order of messages within a conversation?

Now answer it

The same design always gets the same score, with the reason behind every requirement — then defend it against the trade-offs above.