Re: [v6ops] new draft: draft-taylor-v6ops-fragdrop

[Mark Andrews <[e-mail]>]

In message <[e-mail]>, Mark Sm
ith writes:
> Hi,
> 
> 
> 
> ----- Original Message -----
> > From: Merike Kaeo <[e-mail]>
> > To: Lorenzo Colitti <[e-mail]>
> > Cc: V6 Ops <[e-mail]>
> > Sent: Thursday, 25 October 2012 7:49 AM
> > Subject: Re: [v6ops] new draft: draft-taylor-v6ops-fragdrop
> > =
> 
> > =
> 
> > On Oct 23, 2012, at 1:39 AM, Lorenzo Colitti wrote:
> > =
> 
> >>  On Mon, Oct 22, 2012 at 8:44 PM, Nick Hilliard <[e-mail]> wrote:
> >>  So, what do we do?=A0 Do we:
> >> =
> 
> >>  - ignore the problem
> >>  - issue generic advice
> >>  - name and shame
> >>  - other
> >> =
> 
> >>  Other. Document that some networks drop fragments, the reasons why they=
>  do =
> 
> > so, and the impact on applications, but provide no advice.
> > =
> 
> > +1
> >=A0
> 
> I think there should be advice. If the IETF has seen fit to make
> 
> fragmentation a capability of IPv6, then I think that is inherently saying
> that the IETF strongly suggests if not requires forwarding rather than
> dropping of fragments for successful IPv6 operation. This seems to me to
> be an RFC2775 Internet Transparency issue.
> 
> Alternatively, if fragments are more trouble than they're worth,
> they should be deprecated.=A0
> 
> >>  It's unlikely that this group, let alone the IETF, would agree on what=
> =A0
> 
> > advice to give (at least this decade), but I think it is the responsibili=
> ty of =
> 
> > an operations group to document real operational issues, so the rest of t=
> he =
> 
> > community, including application developers and protocol developers, can =
> be made =
> 
> > aware of it.
> >=A0
> 
> It seems to me that developing an application that may be used over the
> 
> Internet involves developing for the worst case, not the best. If a small
> but significant enough portion of sites on the Internet drop fragments, then
> applications will be developed to avoid using fragments. Once applications
> avoid using fragments to suit some sites on the Internet, then all sites
> on the Internet may as well drop them, because they've been rendered useles=
> s.

Or they learn to adapt and use fragments where they work.  DNS clients
do this today.  They get stuck behind equipment that:

	* drops packets > 512 bytes
	* drops fragments
	* drops EDNS packets
	* drops packets with DO set
	* drops packets where the initial fagment doesn't arrive first

The clients adapt the queries they make to get responses back.

They work much more efficiently however if they can get fragmented EDNS
responses back.
 
> > +1 on documenting known issues to make others aware
> > =
> 
> > - merike
> > _______________________________________________
> > v6ops mailing list
> > [e-mail]
> > https://www.ietf.org/mailman/listinfo/v6ops
> > =
> 
> _______________________________________________
> v6ops mailing list
> [e-mail]
> https://www.ietf.org/mailman/listinfo/v6ops
-- 
Mark Andrews, ISC
1 Seymour St., Dundas Valley, NSW 2117, Australia
PHONE: [phone]                 INTERNET: [e-mail]