Date and Time: 6:00am PST / UTC 14:00, Thursday March 03, 2016
Convener: Bin Hu (AT&T)
Georg (Ericsson) presented the project proposal. Some issues of Neutron include:
- Neutron API was not designed with NFV in mind
- Mismatch of Neutron API with certain NFV use cases
- Neutron focuses on L2 in nature
- Strict rules on how to change API
- Nova and Neutron are strongly coupled
- extra complexity and time consuming to introduce new functionality
So this requirement project proposal was introduced. Phased approach will be taken, including collecting use case and requirement, gap analysis, alternative solutions and options, and evaluating the conclusion through prototyping.
The scope of the project covers APIs of OpenStack and SDN Controller. No architecture will be proposed. Solution has to fit OpenStack. Initial use case include L3VPN and SFC.
There are 3 types of concerns from discussion. The first type of concern is the breadth of use cases and unclear milestones and deliverables. Prakash gave the examples that use cases should include edge, access and core networks. And it is not clear what will be done for C release. Do we expect gap analysis for OpenStack Summit in April?
The second type of concern is the alignment with other existing projects, such as SDNVPN and SFC. Are they overlapping in scope? Why cannot we do gap analysis within those existing projects?
Ildiko clarified that the intention of the proposal is not to address those use cases or their functionality (e.g. L3VPN or SFC). Rather, the intention is to support those use cases in an easier and more flexible way by optimizing the design of Neutron API. Bin clarified existing SDNVPN project is to ensure the integration of OpenStack networking-BGPVPN API and ODL backend driver will be properly available on OPNFV platform. It doesn't cover gap analysis and easier way of optimizing Neutron API. The scope and focus is different.
Dave Neary asked if the project will try to make Neutron better or to have another component in OpenStack. Tim Irnich responded that this is what the project needs to find out. Ildiko further indicated that it is not decided yet, whichever best fits OpenStack.
Prakash like to be a contributor for this project and will submit a new use case on Mobile Edge computing, and Ildiko welcomed his proposal and contribution to the project
Patrick asked if there is already collaborative effort with other OpenStack projects. Paul confirmed that there is not such effort yet. Ryota felt that the topic is very important, but the scope seems too large.
The third type of concern is once the prototyping is done, how to handle the code. Georg said that the code will be in OpenStack repo.
After the discussion, it is concluded that three actions are needed:
- Georg will revise the proposal with clear milestone and deliverables at each milestone
- Georg will socialize with SDNVPN, SFC, and test projects to ensure there is alignment and collaboration.
- Georg will add a strategy and plan of how to handle the prototype code if it is not intended in any formal release.