HTTP

What your browser actually says to a server when you open a page

HTTP (HyperText Transfer Protocol) is the format a browser uses to ask a server for something: one request in, one response out. After DNS and a TCP connection, it sends plain text like "GET /news HTTP/1.1" and "Host: example.com".

The first line holds the method (GET fetches, POST sends data such as a form), the path (/news) and the version. Headers follow as "Name: value" lines, such as Host, which names the site wanted.

The server answers with a status line such as "HTTP/1.1 200 OK", its own headers (like "Content-Type: text/html", saying what kind of file follows) and then the body: here, the HTML of the news page.

HTTP

The "404 Not Found" page is a server's HTTP reply

"404 Not Found" is an HTTP status code: DNS worked and TCP connected, but the server has nothing at that path. Codes starting with 2 mean success, 3 a redirect, 4 a bad request and 5 a server failure.

You can watch this yourself: press F12 in Chrome or Firefox and open the Network tab. One news page shows dozens of GET requests, one for the HTML and one per image, script and stylesheet, each with a status code.

HTTP

One open TCP connection carries request after request

The server keeps no memory between requests; HTTP is stateless. That keeps servers simple, so anything to remember, like your login, rides in a Cookie header the browser resends with every request.

The real cost is round trips (RTT: the time for a message to reach the server and a reply to come back). Opening a TCP connection costs one round trip, and each request at least one more.

So HTTP/1.1 keeps the connection open and reuses it for the next file. Requests on it still go one at a time, which is why Chrome opens up to 6 connections per site.