Why does erlang use random ports for distribution

When nodes connect to each other on Erlang, why not just use the original port the connection is made on? Why cau开发者_Python百科se the firewall issues that using random ports causes. I know how to get around this, but I don't understand why.

Edit: I realize that this is frequently done, and this isn't necessarily an Erlang question, but it was a design choice made in Erlang, and I'm curious what the advantage of this method is.

---

It's not abnormal for classic RPC services wanting do ad-hoc communication.

You can specify the ports with kernel parameters, e.g.:

-kernel inet_dist_listen_min 21100 inet_dist_listen_max 21199