Your app went viral

Twitter clone · Level 20 intuition points

Step 1 of 3: 1Hold the load
Step 2 of 3: 2Survive a crash
Step 3 of 3: 3Defend it

Your Twitter clone runs on one $12 server. A tweet went viral, and traffic is heading to 5,000 users. Change the design so the app stays fast and does not fail, without spending too much.

Set 1 of 5

Test suites

not run yet

The design is validok so farnot run yet

Two or more app servers need a load balancer.

Costs $100 a month or lessok so far, now $12 a monthnot run yet

Your design costs $12 a month now.

Holds 5,000 usersp95 under 500 ms, errors under 1%not run yet

Not tested yet. Press Run load test.

Still holds when a server crashes at 60 slocked until the normal test passesnot run yet

Unlocks after your design passes the normal test.

Traffic

Users at the peak: 5,000
Each user sends: 1 request every 5 seconds
Peak requests per second: you work this out

Requirements

p95 latency (slowest 5% of responses): under 500 ms
Errors: under 1%
Budget: $100 a month

What one request costs

About 3 ms of app CPU and 1 ms of database CPU per request.
Node.js uses one core per process. A box with more cores than Node processes leaves cores idle.
One core has 1,000 ms of CPU time every second.
If the database shares App Server 1, its 1 ms is added to that box. On its own server it is free of the app's work.

Prices per month: 1 vCPU $12, 2 vCPU $24, 4 vCPU $48, 8 vCPU $96, 16 vCPU $192. Load balancer $12. A database on its own server costs the same as an app server of that size.

Take it as a Mock

An AI interviewer asks you about this problem, checks your numbers and tests your ideas in the simulator. About 25 to 30 minutes. You get a report.

Uses one AI session from your plan

Numbers are provisional estimates.

Your design

5,000×1$12DB

Users

5,000 users

at the peak

Users grow from 0 to 5,000 in 120 seconds, then stay for 60 seconds. The replay takes about 10 seconds.

Cost$12/ $100 a month

Load test

Run the load test to see the numbers and charts here.

Latency (response time)