[address-policy-wg] Proposal for change to the IPv4 PI allocation policy

Hi,

I warn all my clients that PI (and any kind of their own) address space
and/or AS is also a great responsibility. And they sure need a more
qualified system administrator to rule it.

If client asks for a help ("the Internet doesn't work") and rejects in
help with doing some tests from its side - I say I can't help you in
that case, goodbye. It is true as well for PI as for PA.

Gert Doering wrote:
> Hi,
> 
> On Wed, Aug 23, 2006 at 11:57:39PM +0400, Max Tulyev wrote:
>> Gert Doering wrote:
>>> With PA, you can debug from *your* network - with customer's PI, you
>>> can't (normally) run probes (traceroute, ping) from *their* IP addresses
>>> - and if your PA works, but their PI is filtered, this makes it harder
>>> to debug.
>> PI means that it is *not* your network administratively and technically.
>> If customer have PI and AS - it is an independent part of Internet, and,
>> in fact, not your just customer in usual terms. You don't go debug for
>> example  your downstream's and peer's networks as your, isn't it?
> 
> You need to distinguish between "customers with their own network block,
> their own AS number, BGP speaking routers, multiple upstreams, and local
> clue" and "customers that want PI space because they don't want to 
> renumber when they change upstreams".
> 
> The latter sort will want *you* to announce their PI for them, and if
> there are any routing problems, yell at you because "the Internet doesn't
> work".
> 
> I have no issues with the former sort, especially when there is enough
> local clue present.
> 
>>> Furthermore, /24s tend to be damped more quickly and for longer times
>>> than /16...
>> Again, why? ;)
> 
> Because flap dampening recommendations said so...
> 
> Gert Doering
>         -- NetMaster


-- 
WBR,
Max Tulyev (MT6561-RIPE, 2:463/253 at FIDO)