Le 2012-04-02 à 12:33, Ole Trøan a écrit :

>> If this is to say that until a BOF is started, you will keep your 
>> objection(s) unknown, I continue to take it as a lack of identified 
>> objections.
> 
> the objections I'm aware of are:
> - people are uncomfortable with only a double translation solution

Are you furtively suggesting that 4rd-U has all limitations of a real double 
translation solution like MAP-T? 
That mays sound tactically smart, but isn't justified by facts.


>   * injection of IPv4 routes into IPv6

No need for this with 4rd-U (and BTW not even needed with MAP-T, AFAIK)

>   * transparency

4rd-U is fully transparent.
It has been precisely designed for this  (and to simultaneously keep 
compatibility with IPv6-only O&M).

>   * hard to identify IPv4 traffic in the "core"

The V octet is precisely what facilitates IPv4-traffic identification in the 
core.

>   * higher risk of leakage. i.e single translation / packets outside the 
> domain
>   (all these apply equally to MAP-T btw)

???
If you have a reference explaining which new risk you see, I will look at it.

> - redefining the modified EUI-64 format (V-octet)

As said in the draft, the u=g=1 combinations of the V octet 'differs from "u" = 
0, reserved for local-scope Interface IDs, and also differs from "u" = 1 and 
"g"= 0, reserved for unicast Interface IDs using the EUI-64 format.'
There is therefore no "redefinition" of the modified EUI-64 format.

> and the consequences standardization wise of that
> - overloading information in the fragment header

Using, between CEs and BRs, unused bits of packet IDs doesn't interfere with 
any existing RFC, independently from its being considered overloading or not. 


> (please don't respond to these, FYI only. I'm on Easter holiday.)

Asking that a list of objections shouldn't be answered is to me very unusual!
(You need not to look at my response before you return from holiday!)


> if it wasn't for my general "uncomfortableness" with double-translation I'd 
> could be convinced that a single solution was possible.

> 
> the status quo; with no path forward just means that we'll effectively kill 
> A+P.

Validating 4rd-U with running code, clearly isn't status quo;
It is a path forward which, in case of success, will also be success of the 
generic A+P approach. 


> I would certainly not recommend my product managers to implement either of 
> this given the risk.

Either of this???

> is there consensus to abandon these efforts (which is basically what we do by 
> publishing them as experimental anyway)?

Suggesting that validating 4rd-U would be abandoning the A+P effort is very 
daring. 
It is precisely a way to try and have a clean and unique standard for shared 
addresses:
- working on mesh topologies
- without per-customer state in ISP nodes

> the alternatives we have are perfectly fine:
> - Shared IPv4 address over IPv4 transport -> NAT444 / CGN
> - Shared IPv4 address over IPv6 transport -> 464XLAT / DS-lite
> - Full IPv4 address over IPv6 transport -> DS-lite with Public IPv4 address

Your suggesting that these three might be sufficient is really strange: 
- none of these supports mesh topologies
- In order to support shared addresses, all these need per-customer or per 
session states in network devices.

> problem solved. ;-)

Enjoy you vacation.
RD



> 
> cheers,
> Ole
> 

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

Reply via email to