3 ms·
You get to use 16 bits of the address to chop up hierarchically for your site. You get a /48 and each of your networks gets a /64, with hosts picking random 12
by gorgoiler 4y ago
You get to use 16 bits of the address to chop up hierarchically for your site.
You get a /48 and each of your networks gets a /64, with hosts picking random 128 bit addresses in each.
The 16 bits in between the two subnets mean you have room to breathe for doing whatever you like. Maybe 64k VLANs, or maybe a hierarchy with semantic meaning. You don’t need an IPAM tool if the addresses have meaning.
My favourite: you can route a /56 to a Docker host and have 256 separate bridged networks, all globally routeable. You never need anything like that, but the open space is refreshing. Like I said: room to breathe.
- clord 4y agoAlso in this line of thinking: hosts can be assigned an entire subnet, and applications can get individual /128s. This way, a single host can provide a bunch of independent services, which can be broken out into real machines as the system grows without renumbering.
- gorgoiler 4y agoI think the more important use case is tempaddrs. You don’t have to use the same address all the time, or even at the same time. You can just make up random addresses for each connection if you like, though in practice the rotation is much slower.
- Dylan16807 4y agoIs there a reason to give each host a subnet in that scenario? You don't need to do that just to let a host grab 50 IPs. Actually, if you want to avoid renumbering, don't you want to have that whole block of servers share a subnet?