The history behind a web page arriving

From typing a URL to seeing the page takes a few seconds. In that time your data travels wrapped in envelope after envelope. Every piece of that machinery has a story of how it was born. The site’s “How a Web Page Arrives” is a simplified educational model. Real communication is more complex.

The journey in brief (the site’s “How a Web Page Arrives”)

This work shows what happens until https://example.com/ opens, in 24 steps. The flow is as follows.

1. DNS looks up the address (an IP address) for the name (example.com).
2. TCP builds a connection with three exchanges: “May we connect?”, “Yes, ready?”, “Ready. Let’s begin.”
3. Encryption (TLS) is set up and HTTP asks, “Please send the page.”
4. The data is wrapped in envelopes: encryption, TCP, IP, then Ethernet.
5. It passes through the router, crosses the internet, and the server opens the envelopes in reverse order.
6. The reply comes back the same way and the browser shows the page.

This site’s program checked that the TCP handshake (SYN, SYN-ACK, ACK) comes before the encrypted HTTP, that envelopes are added one at a time and removed in reverse order, and that the Ethernet addresses change at every hop while the IP addresses and TCP ports stay the same to the end (network-journey/engine.test.ts).

The invention of the packet

Packet switching, sending big data in small pieces, was born in the 1960s. Paul Baran of the RAND Corporation in the US thought of it in the early 1960s and published the report On Distributed Communications in 1964. Donald Davies of the UK’s National Physical Laboratory reached a similar idea independently around 1965. The word “packet” caught on from Davies’s choice. Baran called the pieces “message blocks”.

It is true that Baran had the survival of military communications in mind. But “the internet was built to survive a nuclear war” is a claim to handle with care.

The first message was “LO”

On the night of 29 October 1969, Charley Kline, a student programmer at UCLA, tried to log in to a computer at the Stanford Research Institute from Leonard Kleinrock’s lab. After he sent the two letters “L” and “O”, the system crashed. About an hour later the whole login went through. This was the first computer-to-computer message on the ARPANET, the forerunner of the internet.

TCP/IP and “layers”

Vint Cerf and Bob Kahn published a protocol for connecting many packet networks in a paper of May 1974. It was the origin of TCP. It was later split into TCP and IP, and the ARPANET switched to TCP/IP in 1983.

Communication is thought of as jobs divided into stacked layers. The OSI reference model is often described with seven layers, and TCP/IP with four (application, transport, internet and link). The site’s diagram shows five envelopes: HTTP, TLS, TCP, IP and Ethernet. That is neither OSI nor the four layers of TCP/IP; it is this site’s own way of showing it. Where TLS belongs also differs between textbooks. It is not a matter of which is right; there are several ways of looking.

DNS: from names to addresses

DNS is said to have been designed by Paul Mockapetris. Its present specification rests on RFC 1034 of November 1987. Before DNS, every networked computer fetched a single file called HOSTS.TXT from an organization called the NIC. As computers multiplied, that traffic grew too large to cope with. DNS, spread over the whole world, was made as the answer, the account goes. Looking up a number from a name is often compared to a phone book.

The birth of the Web

Tim Berners-Lee wrote the first proposal for the Web at CERN in Switzerland in March 1989. In November 1990 he and Robert Cailliau made a formal proposal, and by the end of 1990 they had built the first browser and server and settled the basic ideas of HTML, HTTP and URLs. It became available inside CERN in March 1991 and was announced to newsgroups in August of that year.

The Web and the internet are different things. The Web is one of the systems that run on top of the internet.

Where the model differs

• The length of the trip: real communication has far finer exchanges than the 24 steps here. TLS, for example, is folded into a single step.
• A little history: the first ARPANET message was “LO”, and before DNS there was a single file.

“How a Web Page Arrives” goes one step at a time. Watch the envelopes pile up and then come off in order.