All articles
Fundamentals 11 min readJune 13, 2026

How the internet actually moves your data

You click a link and a page appears. Here is the whole hidden journey, in plain language.

You type an address, hit enter, and a fraction of a second later a page is on your screen. In that fraction of a second, your request crossed continents, passed through dozens of machines owned by strangers, and came back, all without losing a single letter. Here is how, with no prior knowledge assumed.

Everything is cut into packets

The first surprise: your data does not travel as one piece. A web page, an email, a video, anything, is chopped into thousands of small chunks called packets. Think of mailing a long book by tearing out the pages, putting each in its own envelope, and numbering them. Each envelope finds its own way across the country. At the other end, the pages are put back in order using the numbers.

This sounds inefficient, but it is the secret to the internet's resilience. If one road is blocked, only the envelopes on that road are delayed; the rest reroute. No single failure stops the whole message.

Your message
a page, file, video
Split
into packets
Travel
each its own path
Reassemble
by number
A message is split into numbered packets that travel independently and are reassembled in order at the destination.

Addresses: how a packet knows where to go

Every device on the internet has a number called an IP address, like a postal address for computers (for example, 142.250.72.14). Every packet carries the sender's and the receiver's IP address, so any machine along the way knows which direction to pass it.

But you do not type numbers, you type names like stratiflux.com. Computers need the number. So the internet has a phonebook.

DNS: the internet's phonebook

DNS (the Domain Name System) turns a human name into a machine number. Before your browser can talk to a site, it quietly asks, “what is the IP address for this name?” The answer comes back, usually in milliseconds, and is remembered for a while so it does not have to ask again.

  1. 1
    Ask the resolver
    Your device asks a nearby DNS resolver (usually run by your ISP or a service like Cloudflare) for the address of the name.
  2. 2
    Walk the tree
    If nobody nearby knows, the resolver asks the authoritative servers: first who runs ".com", then who runs "stratiflux.com".
  3. 3
    Get the number
    The authoritative server returns the IP address for the name.
  4. 4
    Remember it
    The answer is cached for minutes to hours, so the next visit skips the whole walk.
Resolving a name to an address. Each layer remembers the answer for a while (caching), so most lookups are nearly instant.

Routing: finding a path through a network of networks

The internet is not one network. It is hundreds of thousands of separate networks (your ISP, a university, a cloud provider) that agree to pass each other's traffic. The word “internet” literally means inter-network.

Special machines called routers sit at the borders and decide, packet by packet, which neighbor to hand it to so it gets closer to its destination. They share maps of who can reach whom using a protocol called BGP (the Border Gateway Protocol). BGP is essentially the internet's GPS: a constantly-updating set of directions between networks. When you hear that “the internet had an outage,” it is often a BGP mistake, one network announced a wrong map and traffic drove into a wall.

The key idea
No one is in charge of the whole route. Each router only knows the next good hop. The full path emerges from thousands of independent local decisions, which is exactly why the internet keeps working even as pieces of it fail.

TCP: making an unreliable network reliable

Packets can arrive out of order, get duplicated, or vanish. So how does a file arrive perfectly intact? A layer called TCP (the Transmission Control Protocol) sits on top and cleans up the mess:

  • It numbers every packet so the receiver can reassemble them in order.
  • The receiver sends back small acknowledgements (“got 1 through 40”). Anything not acknowledged is sent again.
  • It slows down when the network is congested and speeds up when it is clear, so it shares the road politely.

Its faster, no-guarantees cousin, UDP, skips the bookkeeping. Live video and games use it: a dropped frame from a moment ago is not worth resending, just keep going.

TLS: sealing the envelope

On its own, a packet is a postcard, anyone handling it can read it. TLS (the thing that puts the padlock and the “https” in your address bar) seals it. Before any real data flows, your browser and the server do a quick handshake: they verify the server's identity with a certificate and agree on a secret key. After that, everything is encrypted, so the networks in between can carry your data but not read it.

CDNs: keeping a copy close to you

Distance costs time. A server in Virginia answering someone in Lagos has to send packets across an ocean and back, every time. A CDN (Content Delivery Network) fixes this by keeping copies of a site in hundreds of cities. When you load a page, you are usually served from a machine a few milliseconds away, not from the origin. This is why big sites feel instant almost everywhere, and a smart use of one is one of the cheapest performance wins there is.

Putting it all together

So here is the full journey when you open a page, all of it happening in well under a second:

The round trip for a single page load. Most of these hops are cached or kept close, which is why it feels instant.
You seea web page (HTTP)
Sealedencryption (TLS)
Made reliableordering + retries (TCP)
Addressedsource/destination (IP)
Carriedfiber, wifi, undersea cables
The same trip seen as layers. Each layer only worries about its own job and trusts the one below it. This separation is why the internet could evolve for 50 years without being rebuilt.

Why this still matters in ten years

The hardware will change, faster fiber, more satellites, new chips, but this layered design is not going anywhere. It is too good. Every new thing, from video calls to AI agents talking to each other, is built on these same five ideas: cut it into packets, address them, route them, make them reliable, seal them. Understand these once and you understand the plumbing under everything you will ever build.

The one thing to remember
The internet is reliable because its pieces are unreliable and independent. Nothing trusts anything to be perfect; every layer assumes the one below it will sometimes fail and plans for it. That is also a good way to design software.

Written by the Stratiflux engineering team

We build and run this kind of infrastructure and AI for companies, and train the engineers who do it. If a piece of this is on your plate, we can help.

Deep Dive Locked

Enter your email to instantly claim 10 free credits and read the rest of this highly technical deep dive.