Every YouTube video, WhatsApp message, and Google search crosses thousands of kilometres of infrastructure in under 200 milliseconds — here's the exact chain of events.
What's actually happening here?
Think of the internet like a global postal system — except instead of days, letters arrive in milliseconds. Your device is one address, a server somewhere in the world is another address, and a network of cables, routers, and data centres is the delivery system in between. Every time you open an app, your device sends a tiny message asking for data, and a distant computer sends it back — thousands of times per second.
The problem this solves
Before the internet, computers could only share information when they were physically connected to the same machine. The internet solves the problem of getting any computer to talk to any other computer, anywhere on earth, reliably — even when parts of the network break down.
How it really works (step by step)
You type a URL or tap an app. Your device prepares a request — a small packet of data that includes what you want and your device's address (IP address).
Your device asks DNS for directions. The internet uses numbered addresses (like 142.250.80.46), not names. DNS is the phonebook that converts "google.com" into the right number. Your device checks its own memory first, then asks a nearby DNS resolver server.
The request leaves your home. It travels through your Wi-Fi router, then your ISP (internet service provider — like Jio or Airtel), which routes it toward the destination.
Routers pass the packet along. Across undersea cables, fibre lines, and wireless towers, dozens of routers read the destination address and forward the packet — like a relay race. Each router only needs to know the next step, not the entire route.
The destination server receives your request. A server — a powerful computer running 24/7 — processes what you asked for (a page, a video, a search result) and sends the data back in small chunks called packets.
Packets reassemble on your device. Your browser or app collects all the packets, puts them in the right order, and displays the result. This entire round trip often takes under 50 milliseconds.
Here's the full request journey from your device to a server and back:
The part most tutorials skip
Every packet you send takes a different physical path to the server — one might go through Singapore, another through London — and they all arrive out of order. Your device quietly reassembles them using sequence numbers baked into each packet. This is why streaming video can survive a brief network blip: your device has a small buffer of future packets already waiting, so you never notice the gap.
Real company doing this right now
Netflix uses a technology called Open Connect — they place their own servers inside ISPs (including Jio and Airtel in India). When you press play, your video doesn't travel from the US to India; it comes from a Netflix server sitting in your ISP's data centre, sometimes just a few kilometres away. That's why Netflix rarely buffers even during peak hours — they've shortened the internet for their traffic.
Here's the difference between a naive setup and what production systems like Netflix actually do:
What breaks at scale?
If every user in India hits the same single server in the US, you're adding 150–200ms of latency just from the physical distance — that's before the server has even started processing. At 10 million concurrent users, a single server melts. This is exactly why every big app runs dozens of servers behind a load balancer and caches copies of content at edge locations worldwide.
The "aha" moment
The internet doesn't move data from A to B — it moves copies of data, and whoever sends the fewest copies the shortest distance wins.
Your practical takeaway
When your app feels slow, first check where your server is hosted relative to your users — a Mumbai-based server for Indian users can cut latency by 60% versus a US-based one.
Never serve static files (images, videos, CSS) from your origin server directly; put them behind a CDN from day one — it's free on most platforms and immediately speeds up every user's first load.
When debugging network issues, use
traceroute(Mac/Linux) ortracert(Windows) to see every router hop between you and a server — the step where latency spikes is where the bottleneck lives.
Stage 1 · Absolute basics — System Design made easy