Date and Time: 6:00am PST / UTC 14:00, Thursday March 10, 2016
Convener: Bin Hu (AT&T)
Georg (Ericsson) introduced the status of his 3 action items:
- 4 Milestones are defined with deliverables. Prototyping activity will be in parallel with use case collection and gap analysis
- There is no overlap with other projects such as SDNVPN and SFC. Those projects are focusing on integrating existing upstream components, which are not in main upstream package, into OPNFV platform to make sure those functionality are available in OPNFV. But they are not going to investigate new use case, perform gap analysis and optimize the way of Neutron networking.
- Prototyping code will be in public repo, e.g. Github's OpenStack repo.
Prakash (Huawei) suggested Edge Computing use case. Agreement to add Edge Computing as the 3rd bullet under Use Case section. Prakash pointed out dependency on Open Edge Computing project. Georg further indicated that the project is open to new use cases, and open to all possible solutions. Gerald (DOCOMO) proposed to have project proposal updated before going for TSC approval, in particular explaining how Edge Computing will fit into the scope of this project.
Prakash also asked how to make the repo public. Georg clarified that the plan is to host the code in public repo, e.g. Github's OpenStack repo. The discussion went on regarding which public repo should be used. Eventually everyone agreed that as long as it is public repo, the concern is addressed. When project starts, we can discuss the details of which public repo.
When the group was discussing Milestone 4, a question was the potential blueprint. Someone said that doing Blueprint is not OPNFV's job. Then it was clarified that the result could be blueprint or something else. Bin suggested to add to Milestone 4 the recommendation of how to solve the gap in upstream, and drive the solution in upstream if needed.
Ryota (NEC) indicated that it is very confusing, because it includes network everything. Georg indicated that it intends to find a flexible framework to enable Neutron evolution, but not sure if it should be a new component of OpenStack or just Neutron extension.
Dave Neary (RedHat) questioned whether or not we need a new project for it. It seems overlapping with other projects. Tim (Ericsson) said that he is PTL of SDNVPN, and SDNVPN is targeted to integrate upstream component, which is not a main component in upstream package. They don't intend to do use case or gap analysis or Neutron evolution. Dave disagreed with PTL, because they can do some parts of it if they wanted to. But it is PTL's choice if they want to focus on what they are good at. Georg further added that the proposal is to make Neutron ready for NFV use case. Ildiko indicated that Neutron is not created for NFV, so we need to find the gap so that Neutron can be optimized for NFV.
Dave asked if this proposal is similar to ovs4nfv and kvm4nfv. Georg said that yes, and purpose is to make Neutron ready for NFV. The intention is to find the gap. Dave said if that's the case, it is ok.
There was another question regarding how to evolve Neutron. Georg said that it can only be answered after the project is completed. And we expect to find the solution of how to evolve Neutron through this project, i.e. collect use case, do gap analysis, and prototyping work.
Julien asked why we need to do L3VPN within data center. Bin suggested that this type of discussion of use case should be done when project starts to collect use case. It was also clarified that those bullets are examples only and not an exhaustive list in the sense of a scope definition. The selection of use cases to study in order to derive the relevant set of requirements from is what the first phase of the project aims at, So it is not reasonable to attempt concluding this before the project starts.
The meeting ended at 7am PST automatically.