Hello all,

I am working on a trade-off analysis of IPv4 address sharing mechanisms. I am 
replying to Remi's e-mail (sent before I subscribed to the list):

> I therefore worked out a way to present the range of solutions to be 
> compared, 
> with the following taken in consideration:
> - The stateless/stateful IPv4 across IPv6 comparison isn't limited to IPv4 
>   shared addresses (applies also to exclusive IPv4 customer addresses).
> - If there is, in BR/AFBR's, no Customer state (i.e. no states referring to 
>   individual IPv6 prefixes), there can't be per-transport-connection state 
> either.
> - CE-CE direct paths are possible only if IPv4/IPv6 mappings of BR/AFBR's 
> don't 
>   depend on Customer state.
>
> The proposed document structure is as follows, with pros and cons for each 
> section:
> a) Stateful per transport connection (and also stateful per customer IPv6 
> prefix)
>    e.g. DS-lite with CGN
> b) Stateful per customer IPv6 prefix (but Stateless per transport connection)
>    e.g. draft-cui-softwire-host-4over6-06
> c) Stateless per customer IPv6 prefix (and also stateless per transport 
> connection)
>   - Hub-an-spoke
>      TBD
>    - Direct CE-CE paths (mesh)
>      . Encapsulation based
>        e.g. draft-murakami-softwire-4rd-00 alias
>      . Translation based
>        e.g. draft-murakami-softwire-4v6-translation-00

I think we really need such a comparison document. As a newcomer I spent a lot
of time digging up all the scattered documents here and it wasn't easy grasp 
the 
"big picture" of what has been proposed. I think we also need a trade-off 
analysis 
document, which I am writing at the moment.

Anyway, I am not sure I agree with the structure suggested. I think we should 
try
to be even more general and try to abstract all packet manipulations. I have 
prepared
a table which tries to document all so-far implemented combinations (IPv4 
address sharing
mechanisms + 4via6 mechanisms, the intersection includes all but NAT444, which 
is
only IPv4 address sharing mechanism and Public 4over6 which is only a 4via6 
mechanism). 
It would be useful to know if any of the missing combinations might be useful 
in 
practice as well, I hope this table will help in determining that.

I will be very happy to hear any comments on the comparison. It's only a 1-page 
table.

The PDF document: http://nejc.skoberne.net/transfer/mechanisms.pdf

Thanks,
Nejc


_______________________________________________
Softwires mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/softwires

Reply via email to