Hi Ron,

Clearly isolation must not be undersood in strict terms, otherwise
fully isolating the control from the data-plane results in a useless
architecture. This is also what I understand from your comment.

This means that "isolation" must be understood as a degree, "to which
degree are the control and the data-plane isolated?" As such, a
"degree" is not fundamental to an architecture and -in my view- not a
design principle.

Albert



On Sun, Oct 12, 2014 at 12:54 AM, Ronald Bonica <[email protected]> wrote:
> Folks,
>
> Thanks for the good dialog regarding "decoupling" and "isolation". So far, I 
> glean the following from the email thread:
>
> - LISP decouples the forwarding and control plane, so 
> draft-ietf-lisp-introduction is correct
> - LISP does not isolate the control plane from the forwarding plane, so RFC 
> 6830 is also correct
>
> Because both statements are correct and architecturally significant, they 
> *both* should appear in Section 2.1 of draft-ietf-lisp-introduction. Ideally, 
> these two statements should be juxtaposed to one another in order to 
> highlight the difference between "isolation" and "decoupling".
>
> Each statement should include:
>         1)  A title (i.e., Decoupled data and control-plane, Non-isolation 
> between data and control plane)
>         2)  A sentence or two explaining what it means to be decoupled or 
> non-isolated
>         3) A cost/benefit statement
>
>                                                                               
>         Ron
>
>
>> -----Original Message-----
>> From: lisp [mailto:[email protected]] On Behalf Of Darrel Lewis
>> (darlewis)
>> Sent: Wednesday, October 08, 2014 2:07 PM
>> To: Alberto Rodriguez-Natal
>> Cc: Damien Saucez; [email protected]
>> Subject: Re: [lisp] I-D Action: draft-ietf-lisp-introduction-05.txt - 
>> Decoupling
>>
>>
>> On Oct 8, 2014, at 1:34 AM, Alberto Rodriguez-Natal <[email protected]>
>> wrote:
>>
>> >
>> >
>> > On Wed, Oct 8, 2014 at 10:28 AM, Luigi Iannone <[email protected]> wrote:
>> > Hi,
>> >
>> > I think is fair to state in the intro document that data- and control- 
>> > planes
>> are "decoupled" in LISP because their instantiation may run on different
>> boxes, but they are not "isolated" because LISP data plane can trigger 
>> control
>> plane activity.
>> >
>> > I think this is an excellent way to describe it.
>>
>> Agreed.
>>
>> -Darrel
>>
>>
>> >
>> > Alberto
>> >
>> > _______________________________________________
>> > lisp mailing list
>> > [email protected]
>> > https://www.ietf.org/mailman/listinfo/lisp
>>
>> _______________________________________________
>> lisp mailing list
>> [email protected]
>> https://www.ietf.org/mailman/listinfo/lisp
>
> _______________________________________________
> lisp mailing list
> [email protected]
> https://www.ietf.org/mailman/listinfo/lisp

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

Reply via email to