Re: [Int-area] Is IPv6 End-to-End? R.I.P. Architecture? (Fwd: Errata #5933 for RFC8200)

[Joseph Touch <touch@xxxxxxxxxxxxxx>]

On Feb 27, 2020, at 3:54 PM, Mark Andrews <marka@xxxxxxx> wrote:

...
Encapsulation doesn’t make the packet larger.

Not strictly; it does make the unit of transfer larger. It’s the tunnel ingress’s job to adapt to that, e.g., using source fragmentation at the ingress.

It is no different to
talking a packet from a ethernet and pushing it out a PPP link.  The
Enclosing headers differ in sizes but the IPv6 packet remains the
same size.

We’re talking about the unit of transfer - which does get larger….

The difference with encapsulation, from the non encapsulation cases, is
that PTBs are sent to the encapsulating node which has to synthesis a
PTB to the original (as far as the encapsulating node is concerned)

No, it really doesn’t. I don’t expect to chop packets into 48 byte units just to go over an ATM “hop” in an IP path. The only party who needs to know about the PTB of a tunnel transit is the tunnel ingress. Again, see draft-intarea-tunnels.

Joe