What Happens When A Server Suddenly Crashes
The engineering behind big systems becomes visible only after something crashes.
Modern platforms are designed so users continue shopping, streaming, or making payments even when servers fail in the background. The real challenge is not preventing failure — it is building systems that continue working despite failure.
Read the slides as text
Slide 1
ENGINEERING BEHIND BIG SYSTEMS
SYS.01 / 10
Engineering
Engineering
Behind
Behind Big
Big
Systems
Systems
How Amazon Keeps You Logged In
Even After A Server Crash 👀
You are still logged in after a crash
Your cart is safe and untouched
Checkout continues working normally
SYSTEM DESIGN SERIES
Distributed Systems · Backend Engineering
LIVE SYSTEM
USER REQUEST
LOAD BALANCER
SRV-
01
SRV-
02
✕
SRV-
03
SRV-
04
SRV-
05
Slide 2
ENGINEERING BEHIND BIG SYSTEMS
02 / 10
THE PROBLEM
Big Companies Must
Big Companies Must
Solve
Solve This
This Problem
Problem
Amazon runs at planet scale — every single second of every day.
M
Millions
of active users
simultaneously
K
Thousands
of orders placed per
second
$
Non-stop
payments every moment,
24 / 7
ONE SERVER FAILS 💥
✕Users get logged out instantly
✕Shopping carts disappear completely
✕Payments fail mid-transaction
✕The entire website becomes unstable
ENGINEERING BEHIND BIG SYSTEMS
SYSTEM ALERT
COST OF DOWNTIME
For Amazon, even a few minutes of downtime translates to
millions of dollars in lost revenue — and millions of frustrated
customers.
Slide 3
ENGINEERING BEHIND BIG SYSTEMS
03 / 10
ARCHITECTURE
Most People Think
Most People Think
Websites Work
Websites Work Like This
Like This
Many servers working together as one system — that's the
foundation.
ENGINEERING BEHIND BIG SYSTEMS
ARCHITECTURE
NAIVE ARCHITECTURE — DANGEROUS ⚠️
CUSTOMER
→
ONE SERVER
→
DATABASE
Simple. But if that server crashes... everything crashes.
DISTRIBUTED ARCHITECTURE — RESILIENT ✨
CUSTOMER
LOAD BALANCER
SERVER A
SERVER B
SERVER C
SHARED DATABASE
Slide 4
ENGINEERING BEHIND BIG SYSTEMS
04 / 10
THE ANALOGY
Think Of It Like
Think Of It Like
A Shopping
A Shopping Mall
Mall 🏬
🏬
Amazon is not "one counter."
It's a giant mall with multiple counters, multiple staff, and shared
customer information.
ENGINEERING BEHIND BIG SYSTEMS
ANALOGY
01
Counter Active
Serving customers
02
Counter Down
Suddenly closed 💥
03
Counter Active
Instantly takes over
04
Counter Active
Continues serving
KEY INSIGHT
When Counter 02 closes suddenly, another counter continues
serving customers instantly. Customers barely notice
anything happened.
That is exactly how backend systems are designed at scale.
Slide 5
ENGINEERING BEHIND BIG SYSTEMS
05 / 10
LOAD BALANCER
The Traffic
The Traffic Controller
Controller 🚦
🚦
Before your request reaches a server, it first passes through a
Load Balancer — the traffic controller of the internet.
< 1ms
Detection Speed
⚡
Instant Redirect
0
User Impact
ENGINEERING BEHIND BIG SYSTEMS
ROUTING
👤 YOU OPEN AMAZON
⚡ LOAD BALANCER
"Send this user to Server 21"
SRV-21
Active ✓
SRV-23
Active ✓
SRV-24
Active ✓
SRV-22
Crashed ✕
Slide 6
ENGINEERING BEHIND BIG SYSTEMS
06 / 10
SESSION STORAGE
The Secret Behind
The Secret Behind
"Still Logged In"
"Still Logged In" 🔐
🔐
Your login session is NOT stored inside one server only. It lives in
shared systems accessible by every server.
SESSION STORE
Redis Cache
Shared Session DB
Auth Service
EVERY SERVER KNOWS
Who you are
Your login status
Your active session
SERVER A
Crashes 💥
→
SHARED SESSION
Still intact
→
SERVER B
Continues ✅
ENGINEERING BEHIND BIG SYSTEMS
AUTH LAYER
RESULT
No re-login required. No interruption. The session persists
across any server in the cluster — completely transparent to
you.
Slide 7
ENGINEERING BEHIND BIG SYSTEMS
07 / 10
DATA PERSISTENCE
Your Shopping Cart
Your Shopping Cart
Also
Also Survives
Survives 🛒
🛒
You added items to your cart. Then a server crashed. What
happened to your selections?
👟
Shoes
⌚
Watch
🎧
Headphones
IF CART LIVED INSIDE ONE SERVER...
Every item would disappear after a crash. Total data loss.
SERVER A
Crashes 💥
→
CLOUD DATABASE
Cart safely stored here
→
SERVER B
Reads cart ✅
Cart remains exactly as you left it
Wishlist persists across any server
Orders continue normally without interruption
ENGINEERING BEHIND BIG SYSTEMS
DATA LAYER
Slide 8
ENGINEERING BEHIND BIG SYSTEMS
08 / 10
REAL-TIME FLOW
What Happens
What Happens
During A Crash
During A Crash — Live
— Live
Here's the exact sequence behind the scenes, in real time:
ENGINEERING BEHIND BIG SYSTEMS
FAILOVER
1
User is connected to Server A — shopping normally
2
Server A crashes unexpectedly 💥
3
Load Balancer detects failure in milliseconds
4
User silently redirected to Server B
5
Server B checks shared session + cart data
6
Login + Cart instantly restored from shared store
7
User continues shopping — completely unaware
CUSTOMER VIEW
"Nothing happened." 😎
ENGINEERING VIEW
"Failover successful." 🔥
Slide 9
ENGINEERING BEHIND BIG SYSTEMS
09 / 10
INDUSTRY SCALE
Why Big Tech Invests
Why Big Tech Invests
Millions
Millions In This
In This
A
Amazon
E-commerce · AWS Cloud
N
Netflix
300M+ streaming users
U
Uber
Real-time ride matching
C
Careem
MENA mobility leader
These systems are built to meet one standard: always available.
Handle millions of
simultaneous users
Recover automatically after
any failure
Stay online 24 hours a day, 7
days a week
Prevent every second of
customer frustration
ENGINEERING BEHIND BIG SYSTEMS
ENTERPRISE
THE STANDARD
Modern users expect apps to work ALL the time. Backend
engineers make that expectation a reality — quietly, invisibly.
Slide 10
ENGINEERING BEHIND BIG SYSTEMS
10 / 10
FINAL THOUGHT
The Real Power
The Real Power
Of
Of Engineering
Engineering
Most users never notice it.
But backend engineers build the invisible foundation of the
internet. 🚀
If this changed how you see the
apps you use every day
Share this with someone in tech — or someone who thinks
backend engineering is just "pushing buttons" 😄
Follow
Repost
Comment
"
The real power of engineering is not preventing
failure…
It is designing systems that CONTINUE WORKING
even after failure.