On 3/9/19 18:39, Tom Herbert wrote:
> 
> 
> On Tue, Sep 3, 2019, 8:31 AM Fernando Gont <[email protected]
> <mailto:[email protected]>> wrote:
> 
>     On 3/9/19 17:33, Templin (US), Fred L wrote:
>     > Why was this section taken out:
>     > 
>     >> 1.1.  IP-in-IP Tunnels
>     >>     
>     >>    This document acknowledges that in some cases, packets must
>     be     
>     >>    fragmented within IP-in-IP tunnels
>     [I-D.ietf-intarea-tunnels].     
>     >>    Therefore, this document makes no additional recommendations     
>     >>    regarding IP-in-IP tunnels.
>     >
>     > Tunnels always inflate the size of packets to the point that they
>     may exceed
>     > the path MTU even if the original packet is no larger than the
>     path MTU. And,
>     > for IPv6 the only guarantee is 1280. Therefore, in order to
>     robustly support
>     > the minimum IPv6 MTU tunnels MUST employ fragmentation.
> 
>     Isn't that an oxymoron? If fragmentation is fragile, if you need
>     something robust, you need to rely on something else....
> 
> 
> Not really, to say fragmentation is fragile is a subjective statement,
> not a quantifiable fact.

RFC7872. IIRC, Geoff did his own independent measurements.



> As discussed on the list, fragmentation is
> productively in use in many networks that employ tunneling. In those
> cases fragmentation is not fragile and so there's nothing to fix.

The measurements I've seen about fragmentation on the big-I internet
seem to indicate that it is fragile. Yes, you can always build your own
private network where it is not.


-- 
Fernando Gont
SI6 Networks
e-mail: [email protected]
PGP Fingerprint: 6666 31C6 D484 63B2 8FB1 E3C4 AE25 0D55 1D4E 7492




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

Reply via email to