Hi Joe,

> -----Original Message-----
> From: Joe Touch [mailto:[email protected]]
> Sent: Thursday, February 26, 2015 2:11 PM
> To: Templin, Fred L; Ronald Bonica; Carlos Pignataro (cpignata)
> Cc: [email protected]
> Subject: Re: [Int-area] draft-ietf-intarea-gre-ipv6
> 
> 
> 
> On 2/25/2015 4:01 PM, Templin, Fred L wrote:
> ...
> > I have proposed  sending PTB with MTU less than 1280 before as well:
> >
> >    http://www.ietf.org/archive/id/draft-templin-6man-linkadapt-01.txt
> >
> > The idea is to enlist the original source’s aid in easing the frag/reass
> > burdens from the tunnel ingress and egress.
> 
> IMO, PTB is the wrong message for that.

What we have today is a "Type 0" PTB message. What I proposed was a
"Type 1" PTB message.

"Type 0" PTB is the network's way of saying "the message is too big to
deliver", while "Type 1" PTB is the network's way of saying "the message
is not too big to deliver, but it is painful". So, having the tunnel ingress
send a "Type 1" PTB message to the original source is a way to enlist
the source's aid in reducing the pain.

> We don't have a good one to express "preference" vs absolute capability;
> PTB is absolute capability, and I don't agree that it should be abused
> this way.

"Type 0" PTB is about absolute capacity, yes, and so it should not be sent
for sizes smaller than 1280 (because the tunnel ingress is required to pass
at least 1280 even if some fragmentation is needed). "Type 1" PTB is about
preference - sort of like a "quench" message. It is the network's way of
saying to the source: "You are causing me pain which in the long run will
also cause you pain. So, please help ease the pain."

Thanks - Fred
[email protected]

PS A legacy node that mistakes a "Type 1" PTB for a "Type 2" PTB will
    harmlessly begin sending packets as atomic fragments. The tunnel
    ingress can then fragment the payload packet, and an optimization
    will still be realized.

> Joe

_______________________________________________
Int-area mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/int-area

Reply via email to