Search RPD Archives
[rpd] Comments on AFPUB-2018-V6-001-DRAFT01
sander at steffann.nl
Sat Apr 28 15:44:01 UTC 2018
> There’s no reason in IPv6 for non-aligned (nibble boundary) blocks to be handed out. It’s a complete waste of human resources and it makes things more error prone.
> Rather, if we are going to open this particular can of worms, I’d like to see us model things more along the lines of the current ARIN policy:
> 1. Figure out the number of end sites you expect to serve in your largest aggregation point
> in 3-5 years.
> 2. Round that to a nibble boundary (with a 25% minimum free space) (1-12 end sites = 4 bits,
> 13-192 end sites = 8 bits. 193-3,072 end sites = 12 bits, 3,073-49,152 end sites = 16 bits,
> 49,153-786,432 = 20 bits, etc.)… Call this E.
> 3. Figure out the number of aggregation points you expect to have in 3-5 years. Round that up
> to a nibble boundary with a 25% minimum free space (same as in step 2). Call this A.
> 4. 48-(A+E) = prefix size.
> Example: An ISP has 42,000 customers in it’s largest end site. It has 128 end sites.
> E = 16, A = 8, 48-(16+8) = 48-(24) = 24, this ISP should get a /24.
I like the simplicity. I'm not sure if it fits into this policy proposal, but I would definitely support such a proposal if someone would write it. And not only in the Afrinic region ;)
More information about the RPD