Without invoking any analogies, a port is just a number. When an application on your computer sends or receives data, there is a port number associated with it. A server-side application listens for data with a particular port number, and a client side application needs to send data with the same port number to communicate with the right server application. The operating system uses the port number to route incoming data to the right application, and it ensures that only one application at a time can use any given port number.

Some port numbers are assigned to specific protocols (by IANA, I believe), like 80 for HTTP and 443 for HTTPS, so when you see a URL, the default port is usually implied by the protocol, but it can always be specified. For instance, https://google.com is equivalent to https://google.com:443. For more obscure protocols without assigned port numbers, you’ll usually see the port number in a URL, and this tends to happen in the same scenarios where you don’t have a domain name, so you’ll also see an IP address in a URL. It also happens when you need to run more than one of the same kind of server on a single machine. For example, when developing an HTTP server app, it’s customary to use port 8080 or 8888 to distinguish it from the “official” server app on the same machine using port 80, so your development server app will have a URL that looks like http://192.168.0.1:8080.

Typically ports 0-1023 are reserved by the operating system for programs set up by an administrator, and ports starting at 1024 up to a maximum of 65535 are available to any user, so they’re perfect for, say, a Jellyfin server or an app you’re developing. If someone gives you a URL with a port number, especially if it’s above 1023, make sure you trust the owner of the URL, because it can be a giveaway that someone is doing something shady.