Re: [Idr] community of the day

https://tools.ietf.org/html/draft-heitz-idr-large-community-04
as approved by the WG for adoption does not share a common header
with any other BGP attribute. Judging from emails to the list,
there is widespread support for keeping it that way.

The first 7 attributes were added to BGP by RFC 1654 in 1994.
2 are reserved. 24 attributes have been added in the 22 years since.
There are 223 attribute codes left. At that rate we will run out in
over 200 years. And then we can define a common header under which
to define another 200 years of attributes.

I understand that there have been 3 "incidents" in the past,
caused by BGP bugs regarding attribute handling:
 o Corruption of unknown attributes,
 o Attr 128 ATTR_SET handling,
 o leak of confed ASN in AS4-PATH.
This has lead to many ASes filtering all unrecognized or unnecessary
BGP attributes.
RFC7606 will now prevent session resets under these and similar conditions.

If we define a header today under which to group a set of new 
BGP attributes, then if there is a bug in just one of them,
then ASes will filter the common header, thus filtering every
attribute under it.

I think large communities should be a BGP attribute not under
another header.

Thanks,
Jakob.

> -----Original Message-----
> From: Idr [mailto:idr-bounces@ietf.org] On Behalf Of Randy Bush
> Sent: Wednesday, September 14, 2016 4:45 PM
> To: Mikael Abrahamsson <swmike@swm.pp.se>
> Cc: idr@ietf.org
> Subject: Re: [Idr] community of the day - common header
> 
> > Since we don't have that, then now is the time to standardize a "does
> > the same as we have today for 16bitAS but for 32bitAS" which involves
> > as small code, operational and standardization changes as possible, so
> > code can get into operator routers ASAP.
> >
> > So get it out the door now. Not in 6-12 months time. Now. ASAP.
> 
> yes!
> 
> _______________________________________________
> Idr mailing list
> Idr@ietf.org
> https://www.ietf.org/mailman/listinfo/idr