Re: [Ila] [5gangip] Fwd: New Version Notification for draft-herbert-ipv6-prefix-address-privacy-00.txt

[Tom Herbert <[e-mail]>]

On Wed, Feb 21, 2018 at 8:03 AM, Mikael Abrahamsson <[e-mail]> wrote:
> On Wed, 21 Feb 2018, Tom Herbert wrote:
>
>> Host are assigned addresses from the space. The goal is that all addresses
>> appear to be randomly assigned in that space. Conceptually, a host could
>> request assignment for inidivual millions or billions of addresses, but that
>> won't scale. The alternative suggested is to assign addresses in blocks
>> using hidden aggregation as described in section 6.2.2.2. Blocks would
>> contain multiple addresses (could be thousands, millions, etc.).
>
>
> This means the host still needs to cycle these blocks over time to maintain
> privacy. Does the scaling still support a host having hundreds of these
> blocks at any given time?
>

Mikael,

Yes, blocks are a finite resource and so would have to have TTL and be
reclaimed. Also, if keys are involved in address creation they would
need expected management and key rotation. Address space should be big
enough to make long TTLs.

The hidden aggregation method is intended to make scaling possible.
Each assigned block results in on entry in mapping system so total
amount of state is num_hosts*num_blocks per host. e.g. in a network of
10M nodes with 100 blocks per host that's 1B entries in the mapping
system-- should be able to scale that. If block size is 2^16 then if
every address is in the mapping system it's 6.5e11-- harder to scale.
Aggregation is needed keep network state reasonable for scaling, but
aggregation that is visible to the outside world exposes information
about network internals of network and potentially PII as discussed in
the draft.

Tom

>
> --
> Mikael Abrahamsson    email: [e-mail]