RE: Meta-issues: On the deprecation of the fragmentation function

["Templin, Fred L" <[e-mail]>]

Hi Ran,

Cutting a bunch of stuff that basically repeats what I
already said:

> Well, at the moment, on-path fragmentation is not actually prohibited,
> as near as several implementers (including me) can tell, although it
> absolutely is never required to be supported by routers.

Routers that configure RFC2473 tunnels require fragmentation
in order to meet the IPv6 minMTU. This is not fragmentation
of the original packet before encapsulation but rather
fragmentation of the tunneled packet after encapsulation.
 
> Everyone agrees that on-path fragmentation by routers is generally
> undesirable.

Again, never say "always" and never say "never". Undesirable
or not, fragmentation at tunnel routers is unavoidable in
some circumstances.

> Everyone also agrees that a router, for example a backbone high-speed
> router,
> must not be forced to fragment anything -- for practical performance
> reasons
> that are widely understood.

As long as there is an IPv6 minMTU, tunnels will need to support
some form of fragmentation. And, because PMTUD doesn't work very
well, there needs to be an IPv6 minMTU below which the host is
guaranteed that its packets will be delivered. 

> This difference between "not required" and "prohibited" is critical
> for IPv6 deployment over low-rate links having a smaller MTU size,
> as I outlined in my previous note within this thread.

You seem to have not seen the previous thread where I already
showed the need to accommodate low-end links (6lowpan, aviation,
tactical military, etc.).

> If a goal of this WG is to encourage deployment of IPv6 and transition
> from IPv4 to IPv6, then it would be best not to change the current
> status
> of fragmentation in IPv6 -- and it would be helpful to change the
> minimum
> Link MTU to a value lower than 1280 (or deprecate the Minimum Link MTU
> specification entirely, although that seems quite unlikely here now).

I don't think changing the minMTU to something smaller than
1280 now is necessary or even desirable. And, since PMTUD
doesn't work very well, there needs to be *some* minMTU
below which the host can be absolutely assured that its
packets will get through.

> I'd really like to see IPv6 get universally deployed.  These particular
> changes are significant barriers to that happening and create perverse
> incentives either to stay with IPv6 forever or to build & deploy
> IPv6::IPv4 translational gateways (ugh).

I want IPv6 too. But, I want it to go forward with no
limits on migration to larger packets. In other words,
no more magic numbers except 2^16 for IPv4 and 2^32
for IPv6.

Thanks - Fred
[e-mail]