Re: [Idr] WGLC on draft-ietf-idr-as-private-reservation-00

[Christopher Morrow <[e-mail]>]

+idr back

On Wed, Dec 12, 2012 at 2:25 PM, Christopher Morrow
<[e-mail]> wrote:
> On Wed, Dec 12, 2012 at 1:57 PM, Robert Raszuk <[e-mail]> wrote:
>> Chris,
>>
>> Do you expect that folks will use the new space as soon as it is out
>> of RFC editor ? Is there such huge demand for it among ISPs or
>> DSL/FTTH/CABLE operators ;) ?
>>
>
> the use for the ietf-draft-transition space was ... fast. I'm not sure
> how quickly they'll get used, I was simply asking of thought should be
> put into the rollout/phased-approach and what happens in a mixed-node
> network.
>
>> And if so do you really assume those who will use it are so stupid
>> that they would leak those ? I think 4 octet AS itself is not that
>> widely deployed yet. Yes I saw your mail on few two octet private as
>> leaked.
>
> today we see leaks, all the time, of private-as in the path and as
> origin... I clipped far back in this thread one example of many
> examples I found by just asking routeviews.
>
> I have no faith that we'll be successful not leaking this time around.
>
>>
>> And if those are leaked do you think transits should just blindly
>> remove those and replace those private ASes with their own AS ? Note
>
> midstream you can't (today) replace (what would you replace it
> with?)... so no, I don't think the transits will replace, I also don't
> think they'll filter the routes (since they don't today, reliably).
>
>> last time I checked it is not easy to remove private-as from AS_PATH
>> other then replacing it with your own when the UPDATE crossed few
>> hops.
>
> agreed, uniess it's the immediate asn[0] or set of immediate asns[1] I
> think you're out of luck replacing.
>
> -chris
>
> [0]: [] 65543
> [1]: [] 65543 65542 65541
>
>>
>> rgs,
>> r.
>>
>>> On Wed, Dec 12, 2012 at 7:48 PM, Christopher Morrow <[e-mail]> wrote:
>>> On Wed, Dec 12, 2012 at 1:43 PM, Robert Raszuk <[e-mail]> wrote:
>>>> Are you putting this regex in today ?
>>>>
>>>> Otherwise the sooner this goes through your vendors will be able to
>>>> provide you one keyword match for the new private AS range in the
>>>> policy language of your choice.
>>>>
>>>> So I can't quite see what's the point of this regex debate now ....
>>>
>>> to rephrase, I think Robert is asking: "Would you not just ask your
>>> vendor to include a macro like: 'PrivateASNGroup' to match on in
>>> routing policy language?"
>>>
>>> seems reasonable... except that my SFO router may not get upgraded at
>>> the same as my NYC router, nevermind fast-mover Nick over there who's
>>> always happy to load nightly code on his elbonian routers... so do we
>>> have to delay usage of this new shiny space until reasonably close to
>>> all public bgp speakers upgrade? Or build a mechanism like AS4
>>> required for as2 only realms stuck in the middle of the AS4 world?