Hi Ian,


On 15/02/2013 09:29, "[email protected]" <[email protected]> wrote:

>Hi Ole,
>
>Assuming that the Unified CPE draft gets adopted by the workgroup, then
>there needs to be alignment of the different drafts reflecting this.
>
>The unified CPE draft describes how a CPE interprets the presence (or lack
>of) configuration parameters to understand which softwire mode to
>configure. If a MAP CE implemented this (to be a 'unified MAP CE'), then
>it would have two ways of configuring 1:1 mode - via the presence of the
>tunnel endpoint address, an IPv4 address (and optionally a restricted port
>range) and also through a BMR with EA=0.

The unified CPE draft, at this stage, still does not reflect some of the
main feedback provided in :
http://www.ietf.org/mail-archive/web/softwires/current/msg04960.html

Namely: Functional decomposition...

>
>Two ways of configuring the same function doesn't seem like a good idea,
>even if the underlying mechanism that this function is implemented with is
>different.

...And this is the essence of the problem in the "unified CPE" draft. The
underlying implementation MUST not be different, otherwise we're not
unifying anything here.
I will assert that it is possible to have a single implementation that
delivers both MAP and Lw46 characteristics. The proof point of that can
also be found in the existing MAP implementations.

>
>So, what I would propose is that EA=0 for MAP is not included as a
>provisioning option in the MAP draft. If the parameters described in the
>unified CPE draft required for 1:1 mode are configured on the unified MAP
>CE, then it should interpret these to mean EA=0 and configure itself
>accordingly.
>
>
>I also think that it would be cleaner if this 1:1 functionality was
>described in a separate document to the current MAP draft. As the 1:1 mode
>functionality of MAP is a big architectural change to the mesh mode
>function, it really needs a lot more than 2 paragraphs in order to
>describe what it is and how it is used.

It's quite the opposite. A hub&spoke topology is a subset of a mesh, and
MAP covers both. Changing from one mode to the other is actually nothing
more than installing a IPv4 route on the CPE.
In example terms: Each MAP CE is configured with a BMR and a default
route. The BMR conveys the CPE the info about it's port-set to use. Just
with this information the CE can happily operate in hub&spoke mode. When
the BMR is marked with a "use for forwarding" flag, the CE is able to
communicate in mesh mode. Changing from one mode to the other is nothing
but setting a flag in the same BMR.

It is my opinion that we've discussed this 1:1 mode many many times
before, and at each time concluded that a) it is a natural characteristic
of MAP ii) it would actually require *more text* (and complexity) to
remove it.

That said, the biggest take away for me of this thread so far is that the
unified CPE draft seems to be going with the assumption of two
implementations for what is essentially the same functionality. This is a
show-stopper IMO for the unified-cpe draft.

Regards,
Woj..

>
>
>Best regards,
>Ian
>
>On 14/02/2013 11:14, "Ole Troan" <[email protected]> wrote:
>
>>> #25: Maintain or remove MAP1:1 Mode?
>>
>>OK, so here is a task for whomever thinks MAP 1:1 mode should be removed.
>>
>>- what does "remove MAP 1:1 mode mean"?
>>- please suggest text changes to the mechanism that removes 1:1 mode.
>>
>>given that my opinion is that 1:1 mode is an unremovable part of MAP, the
>>question just doesn't make sense to me.
>>
>>I don't want this issue to be an excuse to block a last call, can we
>>quickly resolve this, and can we agree to drop it if there are no
>>significant contributions within the next week?
>>
>>cheers,
>>Ole
>>
>>
>>> 
>>> The WG discussed several times this point (refer to the mailing list
>>> archives).
>>> 
>>> MAP1:1 mode is a particular mode which may re-use some of the
>>>provisioning
>>> methods defined for MAP.
>>> 
>>> MAP1:1 vs. Lw4o6:
>>> * MAP1:1 is not fully stateless.
>>> * Lw4o6 is a standalone specification which provides the same service
>>>as
>>> MAP1:1.
>>> 
>>> -- 
>>> 
>>>-------------------------------------+----------------------------------
>>>-
>>>--
>>> Reporter:                           |      Owner:  draft-ietf-softwire-
>>>  [email protected]       |  [email protected]
>>>     Type:  defect                   |     Status:  new
>>> Priority:  major                    |  Milestone:
>>> Component:  map-e                    |    Version:
>>> Severity:  -                        |   Keywords:
>>> 
>>>-------------------------------------+----------------------------------
>>>-
>>>--
>>> 
>>> Ticket URL: <http://trac.tools.ietf.org/wg/softwire/trac/ticket/25>
>>> softwire <http://tools.ietf.org/softwire/>
>>> 
>>
>>_______________________________________________
>>Softwires mailing list
>>[email protected]
>>https://www.ietf.org/mailman/listinfo/softwires
>

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

Reply via email to