Why is it impossible to contact a home server without a permanent IP, while any server in the world can communicate with a home computer?

Question

Network routing

It is understood that the home address is often the provider's internal address, and that the provider mediates the transmission of the packet.
But - how then do servers from the global network communicate with the internal computer?
And in addition, on some providers, the address received by the router matches the address that determines, for example, 2IP. Why not a white IP?

Answer 1

To connect through NAT (it doesn’t matter if it’s a subscriber’s or a provider’s), if the subscriber does not have a white IP, an external server is used, to which both sides of the exchange establish a connection. And infa goes through it.
If the provider already gives a white IP (only dynamic), you can use some kind of DDNS service and port forwarding on your router.

Answer 2

The source of communication between a PC and global servers is a PC, to ensure the correct return of packets from these servers, NAT technology (or PAT, port-based NAT) is used. When a connection is established, the edge device on which NAT works remembers where and where the request went from, and stores inside itself the IP and port to which response packets from the server should come, and then translates them to the internal network so that they fall on the open socket on the PC that sent the request. And if someone from inside tries to establish a connection with a PC behind nat without a request, this device will not have data that the PC needs to be connected to this server, and the packets will be dropped.
There is a technology that allows you to bypass NAT for direct PC-to-nat communication, but again, the process involves sending requests from this PC to the outside, while the NAT translation is created in a regular way. The technology is called STUN, it works on top of UDP, because UDP does not need to establish a connection via SYN / ACK packets, but you can simply send them.