Hi Yoav,
I understand the adoption practice :-)
What I don't understand is why draft-kumar is titled "Information Model for
Consumer-Facing Interface", as it is not an information model.
In addition, this draft conflicts with the capability I-D, which has
already been adopted. For example, the very first object that is described
is a Policy object, which "represents a mechanism to express a Security
Policy by Security Admin (i.e., I2NSF User) using Consumer-Facing interface
toward Security Controller; the policy would be enforced on an NSF." This
object conflicts with the SecurityECAPolicyRule object defined in the
capability draft.
Especially because there is a normative requirement in the next line of the
kumar draft ("The Policy object SHALL have following information").
Now, if I look at this object, I can make the following comments:
- The information specified is a mixture of attributes and relationships
to other objects, but the actual format and syntax is not specified
- Name and descriptions SHOULD NOT be defined in this spec: they are
inherited from external specs as defined in the Capability draft
- Multi-tenancy SHOULD NOT be defined as an attribute! This appears to be
a set of relationships to other objects
- End-Group SHOULD NOT be defined as an attribute, appears to be
relationships to other objects, and has problems in its definition
- Threat-Feed SHOULD NOT be defined as an attribute, appears to be
relationships to other objects, and has problems in its definition
- Telemetry Data SHOULD NOT be a "field"
- Rules (see below)
- Owner (see below)
What really bothers me is the "Rules" field. This is completely
contradictory to the capability draft. Please re-examine it.
So, if you are asking if I support WG adoption, then the answer is NO.
However, my point was that this draft, besides not being organized as an
information model, is in too incomplete a state to start working on - all I
have is questions.
regards,
john
On Thu, Nov 2, 2017 at 11:39 AM, Yoav Nir <ynir.i...@gmail.com> wrote:
> Hi, John
>
> On 2 Nov 2017, at 7:08, John Strassner <john.sc.strass...@huawei.com>
> wrote:
>
>
> <snip/>
>
> Second, my worry is that draft-kumar is not ready. It is not an
> information model; rather, it is (at best) requirements that could be
> turned into an information model. In addition, it needs to be integrated
> with the existing capability draft. Note that it is absolutely essential
> that we only have a single information model. Having multiple information
> models is akin to having multiple dictionaries; what inevitably happens is
> that the same concept is defined in multiple conflicting ways. I suggest
> that this document be examined in more detail to determine how best to
> proceed. I have already talked to Frank about that.
>
>
> I’d like to point out that a draft that gets adopted does not need to be
> ready for publication. It only needs to be good enough to be a starting
> point for work by the working group.
>
> At present, draft-kumar is the product of its seven authors. They can put
> whatever they want in this document and they don’t need anyone to agree to
> any changes.
>
> What adoption changes is that the group gets change control, so if the
> group decides that the IM should be added in this draft, that is what
> happens; and if the group decides that it should be merged with the
> capabilities draft, that is fine as well. It is not the usual way to have
> a working group work on an individual draft. If we want to work on this, we
> adopt it and make it ready. We don’t wait for individual authors to make
> their draft ready for publications and then adopt it followed immediately
> by working group last call.
>
> I agree that we may want to spend some time on the list of documents
> before adopting them, but getting to start work on the content of these
> documents is what this group is chartered to do.
>
> Yoav
>
>
--
regards,
John
_______________________________________________
I2nsf mailing list
I2nsf@ietf.org
https://www.ietf.org/mailman/listinfo/i2nsf
