Slightly off. Yes double nibble splitting makes sense if you have a 48 or even a 32 but in a customer line 56 is a double nibble aplit from the ISP perspective.
You can do like I do a single nibble split. Like 16 different functions with each 16 sub differentiation.
Are you aware of config option ip6hint to get static subnet identifiers?
You can also if have not done so yet try to request a prefix delegation with the size of 48 and see what happen. But 56 should be fine more home users and even small enterprises hardly hit more then 256 vlans.
I'm not too hot with ipv6. Is it possible to limit the vlan/helper address to 8 bits? I only have five vlans on my home network, so I don't need 12 bits.
I don't think it is causing day-to-day problems because the vlan IDs are simply tagged onto the end of the network address, but it might cause problems if/when a new dynamic ipv6 address is assigned.
With regards to keeping the /64 PD. There's a few threads on Reddit and it seems to be applied to all customers. I did wonder if they've adopted this so that they can assign static IPs to all customers to help mitigate their use of CGNAT.
maybe not the answer you looking for but ...
/56 is ffff:ffff:ffff:ff00:0000:0000:0000:0000
which translated to your "every vlan have own prefix/ID" is 255 vlans
if you want to be honest, what to expect from ISP to give ?
255 vlans for HOME user is more than enough
Thats not the point here. And back in the days there was a proposal that every default connection for an enduser is a 48.
How ever.
I may see the issue here...
You just write the hex-value. So 1 gets 1, and 23 gets 17.
So you can now with your 56 use 0 to FF.
Yes, with your 48 you could have used (like with your ULA) 0 to FFFF.
I don't get it? That's the whole point of ip6hint and udhcpd6c to handle a dynamic prefix delegation and the assignment for the local networks.
here we go
you want to have 1:1 mapping between vlans and v6 prefix, basicaly, prefix=vlan ID
but as i show you, you can not with /56 ffff:ffff:ffff:ff00:0000:0000:0000:0000
simply, you have only 2 byte left for free use, which give you 00-ff = 256 vlans
i am running network in few company arround my place, and even as a company, we simply could not get more than /56
so, i understand the frustration about nice&logical approach with /48 to map vlan ID to v6 prefix, and rest of address space could be assigned to external VPN users, but ...
reality is different
/56 and this is it
Your assumption is still wrong.
Yes a vlan has a range of 12 bit. But if you only have locally 8 bit address space. You can still use all vlans but can only provide global addresses for 256 local networks....
With a 48 assignment every customer could just so participate on the Internet because the customer could manage registry entries by him self.
But then ISP would loose there restrictive monopol.
As I've said, I don't have a problem with the /56 assignment. If they use their additional ISP addresses to provide everybody with a static IP, then all the better.
My question relates to what changes I need to make to my configuration to accommodate the change from /48 to /56.
How many bits does Openwrt consider to be consumed by the ipv6 helper prefix?
As per my screenshots, the subnet is configured to use a /64 assignment length and a helper prefix of 1.
If the helper prefix is stored as a 12 bit value, is that going to cause a conflict due to the overlap with the ISP's network bits?
What I wanted to say is there are only 12 bits of VLAN ID space, so worst case you need 12 bits...
OpenWrt will use anything between the assigned prefix from the ISP and the last 64 bit of the address (called the interface identifier) for its own prefix delegation or what ever assignments you want to use, like your VLAN IDS based approach.
