Hacker Newsnew | past | comments | ask | show | jobs | submitlogin

> Remember, if you allow for identifying servers by ip:port instead of just IP, you have roughly 65536 times as many "addresses" available. That's enough to last pretty much forever.

I see you're using historical precedent:

640k of RAM ought to be enough

32-bit memory addresses will always be large enough

2 digits is enough to encode a year since it will be the 20th century forever

All the characters anyone would want to use fit within 8 bits



You may be right. At that time we will end up introducing or requiring some kind of horrible port multiplexing scheme, like maybe port knocking (using a pattern advertised in DNS) or maybe even... HTTP/1.1's Host: header :)

Separately, I think addresses are slightly different than those other measurements. I'd think the number of needed public server ip:port addresses is roughly on the same order of magnitude as the number of humans, or perhaps less. By that measurement, 4 billion is almost enough, but clearly not enough.

I wouldn't want to have to bet on that, but I don't have to. There's always another layer of indirection possible. And that layer of indirection will always be infinitely easier to deploy than a replacement to IPv4.




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: