fatihyigit.com
MODE
LANG
$500
+$0.0/S
80
+$0 / -$0
100%
0
RPS
0 ms
0.0%
0.05%

SCALE OR DIE

You start with $500, one API server, and one database. Users are coming. If the system slows down they get unhappy and leave; if it stays fast they bring friends. Goal: grow the balance to $1,000,000.

1

Every active user brings in money per second; every resource burns money per second.

2

When capacity fills up, the queue grows, latency spikes, and requests get dropped.

3

Server layer bottleneck: LOAD BALANCER + more servers. Data layer bottleneck: CACHE / REPLICA / QUEUE.

4

As you grow you will face DDoS, data loss, and bad deploys. WAF, BACKUP, MONITORING are how you prepare.

5

If users hit zero or you go bankrupt, the game ends.

SPACE = pause  |  1 / 2 / 3 = speed  |  H = help

SYSTEM MANUAL

REQUEST PATH

USERS > INTERNET > CDN/WAF/RATE LIMIT > LOAD BALANCER > APP SERVER > CACHE / QUEUE / DATABASE

At every layer a utilization rate is computed. Queuing theory means latency doesn't grow linearly as utilization nears 100% - it grows exponentially. At 80% load the system is still fine; at 95% it collapses.

LAYERS

APP SERVER is your request capacity. More than one requires a LOAD BALANCER.

DATABASE is the single write point. REPLICA only helps reads, CACHE stops reads before they arrive, QUEUE makes writes asynchronous.

CDN never lets static traffic reach the origin at all - the cheapest way to scale.

ATTACKS & FAILURES

WAF filters DDoS traffic at the edge. RATE LIMITER cuts bot/scraper load.

BACKUP prevents user loss during a data-loss event. MONITORING shortens every outage.

AUTOSCALER spins up temporary servers under sudden load; expensive, but it saves viral moments.

ECONOMY

Revenue = active users × how happy they are. Cost = the total rent of every resource you own.

Over-buying kills you too: idle capacity drains the treasury. Spend where the bottleneck actually is.

Above 58% satisfaction you grow; below it you shrink.

SYSTEM CRASHED