Re: (ngtrans) I-D ACTION:draft-ietf-ngtrans-mech-v2-01.txt

[Erik Nordmark <Erik.Nordmark@sun.com>]

> 1) does 'tunnel MTU MUST be not exceed 4400 bytes' only apply to the 
> dynamic MTU determination part or also to "manual configuration" part?
> (probably the latter, but it's a bit unclear.)

For manual I assume the encapsulator and decapsualtor should be able to
pick whatever values they want, since this is manual. For instance,
if they have IP over AAL5/ATM in the path they might want to configure a 9k
number.

I take it I need to make this more clear in the draft on this point.

> 2) Why just 4400, not e.g. 4500? (Note: ATM/POS links usually have a
> default MTU of 4470 bytes, so either 4470+20 or 4470-20 (depending on
> which kind of encapsulation you believe is more useful) would be one
> interesting compromise, I believe.)

As I pointed out in emails (but not in the draft) we can debate the actual
numbers - possibly forever. 4400 was something I pulled from a hat.

I don't know what *process* to use to determine the numbers.
(There is also the 1280 default number when the tunnel MTU is not dynamic
which perhaps we need to debate.)

In terms of the 4400 number, it effects a bit the IPv4 reassembly buffer
in decapsulators. This is because the encapsulator might be using 
the dynamic scheme. Thus if a decapsulator have an interface which can receive
packets of size X, it needs a reassembly buffer size for MIN(X, 4400).
Does this mean we can make the 4400 number larger (e.g. 8k, 9k?)

The 1280 number (the MTU used when there is the tunnel MTU is not dynamic)
directly effects the IPv4 reassembly buffer in decapsulators so we
presumably don't want this number to be much larger (but one could argue that
2k would not be a big deal I think).

   Erik