[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index] | [List Home]
Subject: RE: [Sent to NSN.COM] RE: [tosca] Groups - TOSCA-NFV-Profile v1.0 ToC proposal.zip uploaded
Forwarding the email to the Tosca list since Thinh believes there may be email issues with the nsn.com domain.
Hi Chris and John,
I am responding to Chris and John in one email.
I do like your “Hello World” example and build from basic example. Section 3 and section 5 are very much related to give the reader the understanding of TOSCA NFV Profile would layout and works. Perhaps, section 5 should move up before going into normative part of profile Simple YAML in Section 4. Section 4 (TOSCA Modeling Principle and Data Model) is important to begin profile the Simple YAML for NFV. This section is important foundation for defining VNFD and NSD.
The proposal is not to re-write the document. The proposal is to re-factor the document to fulfill the gap such as section 4, and a separation of NSD and VNFD. For the missing items, they are VnfDeploymentFlavour, VnfElementGroup, Vnfindicator, VduCpd, VnfExternalCpd, VnfVld, NsDeploymentFlavour, etc.
I am generally okay reorganizing the sections and providing more info, though I think this will be more work and take more time than some expect.
But I am concerned that we might be starting completely over on technical content that has already been implemented. I would like to see a more pragmatic approach where the changes are more iterative on content that already exists (and has been implemented). For content that is completely new (green field), we should try to introduce it as quickly as possible.
I very much like the direction of this new outline. It takes a big step towards making the document more readable, and it also does a great job in better aligning the Tosca NFV Profile with the abstractions defined by ETSI NFV.
As a potential additional improvement, I would recommend that we go even further and adopt an approach similar to what was done in the Simple Profile for YAML spec. As you know, in that document we built up Tosca concepts by starting from a simple “Hello World” example and gradually adding additional functionality. I believe the NFV Profile document could benefit from a similar build-up.
Specifically, I believe we could use Section 5 in Thinh’s outline for this purpose:
- We would first introduce network service concepts:
o Starting with a very simple example consisting of a single VNF with two external connection points
o Then add a second VNF with a virtual link (and internal connection points) connecting these VNFs
o Potentially adding physical network functions
o Adding forwarding graph(s)
o And any other concepts that might need to be introduced
- Then we could move towards VNF descriptor templates:
o Mapping a VNF onto just one single VDU
o Adding additional VDUs with an internal virtual link
o Specifying how these VDUs use virtual compute and virtual storage
This would make Section 5 the “meat” of the document where most if not all of the NFV concepts are explained and described using TOSCA.
Sections 6 and 7 would then introduce the normative type definitions specified by the document (as shown in Thinh’s outline).
From: firstname.lastname@example.org [mailto:email@example.com]
On Behalf Of Thinh Nguyenphu