[address-policy-wg] Input request for the PI Transfer policy

Hi,

On Tue, Mar 25, 2014 at 02:17:26PM +0200, Elvis Velea wrote:
> On 25/03/14 10:57, Gert Doering wrote:
> > On Tue, Mar 25, 2014 at 10:27:54AM +0200, Elvis Velea wrote:
> >> I, personally, would put the /24 limit in policy. Anything lower than a
> >> /24 can no longer be split by the RIPE NCC and must be transferred in
> >> one block.
> > Why?
> First of all, because the other RIRs (APNIC&ARIN) also have a /24 limit 
> for Inter-RIR transfers and I do hope that our region will eventually 
> have an Inter-RIR transfer policy compatible with theirs .. so, for 
> consistency with the others.

I don't see this as particularily desirable, follow the other regions 
"because they do it this way".  Look there, learn from them, but make
our own decisions.

> Secondly, because most of the routing is done by filtering at the /24 
> level and even though not all the IPv4 space is used on the internet and 
> publicly advertised, it seems to be the limit already adopted by most of 
> the ISPs.

So this should limit the amount of what people *want* to transfer 
automatically, without putting it into the policy, no?  It's commonly
known that announcing a /25 won't work, so why would you transfer one,
if not "for internal use"?

> Thirdly, setting up reverse dns for blocks lower than /24 requires that 
> any minor changes for reverse dns would need to involve the RIPE NCC.

That's indeed quite a strong argument for not *wanting* to transfer 
anything smaller, but would it need to go into the policy document?

(My personal feeling is "nobody would want to transfer smaller blocks
anyway, so we don't need to codify it", but it certainly needs to be
discussed and agreed-upon...)


> Lastly, allowing any level of fragmentation could lead to an exploding 
> IPv4 routing table (now at almost 500k routes). I'm not saying that 
> adding an (arbitrary) transfer limit like the /24 would stop the routing 
> table growth.

It wouldn't... there's about 16 million /24s out there, so even limiting 
at /24 won't stop the explosion if people really start announcing everything
up to the limit...

> There might be other reasons as well.. It's just a feeling that limiting 
> at /24 is the right thing to do. On the other hand, while writing this 
> e-mail I've been having second thoughts on each of the paragraphs and 
> reasons :-)

It's good to have the arguments out, so we can think through them - thanks!

Gert Doering
        -- no hats
-- 
have you enabled IPv6 on something today...?

SpaceNet AG                        Vorstand: Sebastian v. Bomhard
Joseph-Dollinger-Bogen 14          Aufsichtsratsvors.: A. Grundner-Culemann
D-80807 Muenchen                   HRB: 136055 (AG Muenchen)
Tel: +49 (0)89/32356-444           USt-IdNr.: DE813185279
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 811 bytes
Desc: not available
URL: </ripe/mail/archives/address-policy-wg/attachments/20140325/ca41f59b/attachment.sig>