Hi Carsten;

Thanks for the pointer, it's all the more interesting that
tools.ieft.org fails to refer it.
I'll update the fragment recovery draft with the proper reference. 

Now I fail to gather your conclusion: 
Do you mean that applications should refrain from sending more than 80
bytes at a time to avoid 6LoWPAN fragmentation to occur?

Pascal

>-----Original Message-----
>From: Carsten Bormann [mailto:[EMAIL PROTECTED]
>Sent: lundi 16 juin 2008 15:40
>To: Pascal Thubert (pthubert)
>Cc: Carsten Bormann; Geoff Mulligan; Jean Philippe Vasseur (jvasseur);
6lowpan
>Subject: Re: [6lowpan] New charter text: fragment recovery item
>
>On Jun 16 2008, at 10:51, Pascal Thubert (pthubert) wrote:
>
>> http://tools.ietf.org/html/draft-mathis-frag-harmful
>
>Which has since become RFC 4963, "IPv4 Reassembly Errors at High Data
>Rates".
>Mandatory reading in some of our classes (see also RFC 3819, for a
>more general overview of L2/L3 issues), but kind of less relevant for
>IPv6.
>
>(PLPMTUD is RFC 4821, but doesn't really help with the 6lowpann issue,
>because the IPv6 MTU *MUST* be >= 1280, and we would prefer
>applications to stay well below that -- there is no easy way in IPv6
>to tell them.)
>
>The situation right now (RFC 4944) is that applications just have to
>know that large packets will have bad performance.
>This is a classical case of the L2/L3 not being able to tell L4/L7
>something it needs to know.
>Fortunately, many 6lowpan applications will run custom protocols on
>top of UDP, so they can keep this under consideration.
>
>Gruesse, Carsten

_______________________________________________
6lowpan mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/6lowpan

Reply via email to