Search RPD Archives
Limit search to: Subject & Body Subject Author
Sort by:

[AfriNIC-rpd] AfriNIC Policy Proposal: IPv6 ProviderIndependent (PI) Assignment for End-Sites

McTim dogwallah at gmail.com
Tue Mar 27 04:30:34 UTC 2007


On 3/23/07, Colin Alston <colin at thusa.co.za> wrote:
> The fact has been well established that for organizations where (for
> political or other reasons) provider lock-in is unacceptable and where
> LIR status is unobtainable or unnecessary, there is a genuine
> requirement for PI allocations.

I think you mean "assignment", not allocation.  There shouldn't be any
cases where becoming an LIR is unobtainable.  If this is the case,
then we should make it easier for organisations to become an LIR.


 Most organizations are reasonably
> protected from this since they use NAT on IPv4 and use public
> addressing sparingly - switching ISP's is then very easy. IPv6 changes
> this.

How? Why? If ppl like NAT, then they can do that in IPv6!

If you saw how some providers in South Africa treat customers
> with domain names in order to lock them into services, you would be as
> scared as I am about giving them total control over our banks IP
> addressing.

Then use ULA space?

>
> For large organizations who also would like to move faster on IPv6
> uptake than their ISP's (which isn't hard...), it is of great help to
> be able to obtain a PI allocation in order to move forward before
> ability to peer is obtained - again, this goes hand in hand with my
> argument against labor costs of renumbering. In fact by your own
> argument, RFC3177 asserts this reasoning.

Sorry, I can't parse the above para.

>
> You seem to also be of the understanding that any of us actually have
> access to PA space. TENet (which is not a public provider) aside, this
> is unfortunately not a reality for us.

No one will have access to v6 space if no one demands it of their upstream!

> PI should help remove the "chicken and egg" scenario by giving easy
> access to address space to certain content providers and critical
> infrastructures.

If this proposal limited PI to ccTLD's and IXP's I wouldn't be so opposed to it.


ISP's can no longer complain about lack of benefit in
> IPv6 peering on the basis that content density is low if critical
> services are the first to adopt and publicise this fact, the market
> will open drastically. Until that happens, people will continue to sit
> in the dark and work their way around the terrible NAT's already
> inflicting among other things, our mobile providers.
>
> We must try to stop dancing around this issue for once and be brave.

Being brave sometimes means NOT doing the thing that gives short term
gain but long term pain.

The practical issue is where does this PI space come from?

>From yesterday's airrs-report we see:

BGP Report for IP block : 2001:4200::/23 , 2C00::/12
DATE : 2007-03-26
.
.

Minimum allocation size from 2001:4200::/23 :  /32
Minimum prefix size from 2001:4200::/23 in routing table :  /32
.
.
Minimum allocation size from 2C00::/12 :  /32

So does AfriNIC request a new block for this, or do we ask everyone in
the world to change their filters for part of one of the current
blocks?


-- 
Cheers,

McTim
$ whois -h whois.afrinic.net mctim



More information about the RPD mailing list