On Thu, 11 May 2017, Jan Zorz - Go6 wrote:
Dear RIPE IPv6 WG,
As promised at BCOP TF meeting on Monday, the co-authors present at the
RIPE74 meeting gathered on Tuesday afternoon and did some editorial
work, addressing majority of the comments and suggestions we got from
the community based on first version of the draft.
Draft version 2 is now available for reading at
https://sinog.si/docs/draft-IPv6pd-BCOP-v2.pdf
We'll have a short "lightning talk" in Thursday IPv6 WG session, please
go and read the document (those that have enough time and energy), so we
get more feedback and input for further improvements (if needed).
See you all in couple of hours!
Reading this and writing as I read it through:
"IPv6 is not the same as IPv4. In IPv6 you assign a number of “n” /64
prefixes to each end-customer site, so they are able to have as many
subnets as they wish. "
I think this immediately leads the reader wrong. This should be about
sites getting a larger prefix, and THEN out of this, they use /64s. So
while above is technically true, from a viewpoint of making the reader
understand better the hierarchy, I propose above sentence to be:
"IPv6 is not the same as IPv4. In IPv6 you assign a large prefix to each
end-customer site, so they are able to have as many subnets (/64s) as they
need."
The /64 for cellular phones should not be in the executive summary.
4. In IPv4, it's not only perception of scarcity, there *is* scarcity.
4. I think I did the calculation and if you have 8B /48s, you still have
only consumed around 1/10000th of the IPv6 space. I would use this instead
of "480 years". 2^33 is ~8B. 48-33 is 15. So /48 for 8B people uses a /15.
Take that down to one of the /3s we have, and it's a /12. 2^12 is 4096. So
One /48 per person on earth uses 1/4000th of the currently used /3. So
even with inefficient addressing this is not a problem.
4.1.2. Windows PCs CAN do DHCPv6-PD, if they have Internet connection
sharing turned on. This worked already in Windows Vista, 10 years ago.
However, I understand that this is not the point you're trying to make.
4.2.3. Can't we use the "STRONGLY DISCOURAGED" to use less than /56 ?
5.x I have heard of online gamers being ddos:ed so that someone else gains
competitive advantage. It might be good to mention this drawback of
persistent prefixes.
Good document, I'll refer to it a lot because I keep having discussions
with people in different forums about customer prefix size. Thanks!
--
Mikael Abrahamsson email: [email protected]
