Re: draft-bourbaki-6man-classless-ipv6-00

[Brian E Carpenter <[e-mail]>]

On 05/06/2017 02:00, Ca By wrote:
> On Sun, Jun 4, 2017 at 6:10 AM Lorenzo Colitti <[e-mail]> wrote:
> 
>> On Sun, Jun 4, 2017 at 8:05 PM, Philip Homburg <
>> [e-mail]> wrote:
>>
>>> Moving on to a network architecture point of view, when using pseudo
>>> random
>>> IIDs there will be a longest prefix that can be supported. Lets say for
>>> the
>>> sake of argument we can support of /96.
>>>
>>> Then the effect will be that if in the future hosts support SLAAC upto /96
>>> then we are back at the same hard limit. We have just moved be boundary by
>>> 32 bits.
>>>
>>
>> That's *exactly* right.
>>
>>
>>> There is no reason to expect this to change if we make longer prefixes
>>> possible.
>>> It is just that end users will then end up with a /96 and find that they
>>> can't subdivide it any further.
>>>
>>
>> Right. So then someone will say, "we need to extend the network at the
>> edges!". And we move the boundary again, to /112. And then to 120, and then
>> to /124, until we get to one /128 per device. But long before that, a)
>> SLAAC is dead because there's not enough space in the prefix to form a
>> random IID, b) the hosts start doing NAT because they don't want to waste
>> their time. We now have 128-bit IPv4. Except that at least in IPv4, address
>> shortage was real. In IPv6 it isn't.
>>
>> Or, if we remove the boundaries altogether, we end up with /128 straight
>> away, because operational consistency. We now have 128-but IPv4 again.
>>
> 
> Assuming the above path is true ....
> 
> A 64 bit random iid is a very real security advantage.  i will point out
> that things like the recent "wannacry" worm (like many worms before it)
> relied on dense guessable host ip addressing scheme that make address
> scanning effective.
> 
> Making small guessable prefixes increases discoverability and decreases
> security. Please update the security section to include this massive
> regression in security  posture for ipv6 that favors known and commonly
> used network scanning malware techniques of ipv4 will now be used in ipv6
Yes. Any IPv6-over-foo that proposes a boundary at, say, /96 would have to
explain why 32 bits of pseudorandomness is enough. And we could perfectly
well have a guaranteed security DISCUSS on boundaries beyond some
magic number. Maybe that number would even be /64, but I think /80 is
more likely. None of that is the point. The point is to establish
that routing is classless and /64 is a parameter of specific
addressing schemes.

Probably the draft is too long, which obscures that simple message.

     Brian