[ppml] "Recommended Practices" procedure

[Robert E.Seastrom]

Scott Leibrand <sleibrand at internap.com> writes:

> On 06/29/06 at 5:52pm -0400, Marshall Eubanks <tme at multicasttech.com> wrote:
>
>> Hello;
>>
>> On Jun 29, 2006, at 4:53 PM, Scott Leibrand wrote:
>>
>> > I'm sorry, I made an unstated assumption that you're buying transit
>> > from two transit providers who don't have any transit providers of
>> > their own, just peers.  IOW, tier 1 NSPs.  Can you enumerate any
>> > failure modes in that case, or were you just talking about
>> > reachability problems for tier 2 NSPs?
>>
>> I think that any policy that relies on distinguishing between Tier 1 and
>> Tier 2 should be automatically out of bounds. Every time the subject
>> comes up on (say) NANOG it generates much heat but little light, and I
>> know of no tool to enable me to check independently whether or not a
>> salesman's claims here are accurate.
>
> I'm not proposing a policy, I'm proposing text for a "Recommended
> Practices" document.  If we want to eliminate the "distinction" between a
> transit-buying and transit-free transit provider (which IMO is a valid
> technical distinction, egos, tier labels, and salesmen aside), then you
> simply need to expand the recommendation to say something like this:
>
> To multihome with PA space, you must ensure that your transit providers,
> and any of their transit providers (recursive), accept your /48 from each
> other.
>
> Still pretty simple, no?

There's an implicit assumption here that peers (as opposed to upstream
transits) do not filter.  History has shown counting on this to be unwise.

                                        ---Rob