RE: [tosca] NFV: modeling forwarding graphs

[Chris Lauwers <lauwers@ubicity.com>]

Hi Thomas, 

the challenge here is that NFV requires two graphs:

1. A topology graph that specifies how VNFs are interconnected using Virtual Links
2. A forwarding graph that specifies how specific services are constructed on top of this topology

One can think of the forwarding graph as an "overlay" on top of the topology graph.

What you're suggesting makes a lot of sense, and in fact it's consistent with Shitao's proposal for using Tosca to model the topology graph. Shitao introduces Node Types for VNFs, for Virtual Links, and for Connection Points.

What I was trying to address is the second challenge: once we have a VNF topology deployed, how do we model the network service forwarding graph that uses this topology.

Hope this makes sense,

Thanks,

Chris

-----Original Message-----
From: Thomas Spatzier [mailto:thomas.spatzier@de.ibm.com] 
Sent: Tuesday, April 21, 2015 9:03 AM
To: Chris Lauwers
Cc: tosca@lists.oasis-open.org
Subject: Re: [tosca] NFV: modeling forwarding graphs

Hi Chris,

I must admit that I haven't been able to join the NFV ad-hoc group calls due to timing conflicts, so I might miss some background. But I agree to what you said in that using artifacts to describe NFGs would introduce another layer that would need to be standardized (unless the language for the artifact is already standardized e.g. by ETSI).
I agree that it would make much more sense for TOSCA to stick to TOSCA's fundamental modeling concepts of using Nodes and Relationships. In contrast to your proposal, however, I thought more about letting a TOSCA topology
*be* the NFG instead of having something like a NetworkForwardingGraph node type. Basically, one could imaging  node types for VNFs (e.g. VNF1, VNF2,
etc) and relationship types like "ForwardsTo" or "Chains", and then we could have a topology of node templates for the different VNFs with chains relations between them.

... but again the disclaimer that my inability to join WG calls might mean some miss of background.

Regards,
Thomas
---------------------------------------------------------------------------------

Cloud Architect - IT Service Management, IBM Systems
 Master Inventor IBM Deutschland Research & Development GmbH, D229, Schoenaicher Str. 220,
D-71032 Boeblingen, Germany
thomas.spatzier@de.ibm.com | +49-7031-16-1219

IBM Deutschland Research & Development GmbH / Vorsitzende des
Aufsichtsrats: Martina Koederitz
Geschäftsführung: Dirk Wittkopp
Sitz der Gesellschaft: Böblingen / Registergericht: Amtsgericht Stuttgart, HRB 243294



From:	Chris Lauwers <lauwers@ubicity.com>
To:	shitao li <lishitao@huawei.com>
Cc:	"tosca@lists.oasis-open.org" <tosca@lists.oasis-open.org>
Date:	20/04/2015 23:22
Subject:	[tosca] NFV: modeling forwarding graphs
Sent by:	<tosca@lists.oasis-open.org>



Hi Shitao,

You have proposed in the past to use the concept of artifacts to model the forwarding graph(s) in NFV deployments ( https://www.oasis-open.org/apps/org/workgroup/tosca/download.php/54169/Forwarding%20graph%20NFV%20and%20TOSCA.ppt
). However, this would effectively introduce a requirement for another file format that needs to be documented, standardized, and maintained. We would also miss the opportunity to leverage the power of Tosca, since graphs are at the core of what Tosca models are all about.

I’d like to make an alternative proposal where the forwarding graph is modeled explicitly using Tosca Node Types. Specifically, I propose to introduce two new node types: ForwardingGraphNode and ForwardingGraphEdge.
The main responsibility of these node types would be to model the end-to-end graph by connecting ConnectionPoint nodes together:

-          The ForwardingGraphNode connects an “in” connection point to an
“out” connection point on the same VNF (note that “in” and “out” could point to the same connection point)
-          The ForwardingGraphEdge connects a connection point on one VNF
to a connection point on a different VNF.

Tying the ForwardingGraphNode and ForwardingGraphEdge nodes to the appropriate connection points requires the introduction of a new capability type for Connection Points (e.g. tosca.capabilities.nfv.Forwarder) and a new relationship type (e.g. tosca.relationships.nfv.ForwardsThrough) to associate the Graph Nodes and Graph Edges with the appropriate connection points.

Using this approach, we can use Tosca to model both the network topology of VNFs and Virtual Links as well as the forwarding graphs that is overlaid on top of this topology.

Best regards,

Chris

From: tosca@lists.oasis-open.org [mailto:tosca@lists.oasis-open.org] On Behalf Of shitao li
Sent: Wednesday, April 15, 2015 1:59 AM
To: tosca@lists.oasis-open.org
Subject: [tosca] Groups - draft-tosca-nfv-v1.0-wd01-rev02 uploaded



                                                                                        
 Document Name: draft-tosca-nfv-v1.0-wd01-rev02                                         
                                                                                        
 No description provided.                                                               
 Download Latest Revision                                                               
 Public Download Link                                                                   
                                                                                        
 Submitter: Mr. shitao li                                                               
 Group: OASIS Topology and Orchestration Specification for Cloud Applications (TOSCA)   
 TC                                                                                     
 Folder: Working Documents                                                              
 Date submitted: 2015-04-15 01:58:21