Wiki

This page is read only. You can view the source, but not change it. Ask your administrator if you think this is wrong.
==== Minutes of Technical Community Discussion on December 11, 2014 ====

  * ''Date and Time'': UTC 14:00, Thursday December 11, 2014
  * ''Convenor'': Bin Hu (AT&T)
  * ''Participants'':
    * Alexandru Vaduva (ENEA)
    * Andrea Pinnola (Telecom Italia)
    * Bertrand Souville (NTT DoCoMo)
    * Bryan Sullivan (AT&T)
    * Carlos Goncalves (NEC)
    * Chris Price (Ericsson)
    * Cliff Young (ClearPath)
    * Dave Lenrow (HP)
    * Dave Neary (Red Hat)
    * Dirk Kutscher (NEC)
    * Hui Deng (China Mobile)
    * Jonne Soininen (NSN)
    * Juha Oravainen (NSN)
    * Lingli Deng (China Mobile)
    * Luigi Quattrocchi (Telecom Italia)
    * Margaret Chiosi (AT&T)
    * Markus Berglund (NSN)
    * Moiz Arif
    * Parviz Yegani (Juniper)
    * Praveen Kumar
    * Qiao Fu (China Mobile)
    * Rong Gu (China Mobile)
    * Stuart Mackie (Juniper)
    * Tapio Tallgren (NSN)
    * Ulrich Kleber (Huawei)
    * Rajeev Seth (Sonus Networks)

  * Discussion of Integration and Testing project proposals:
    * [[OSCAR/project_proposal|OSCAR]] - a reference platform for validation and testing both open source and vendor VNFs per OPNFV guidelines

Stuart Mackie from Juniper presented this project.

Bryan suggested that other combinations of different components, such as ODL, should be made possible at release 1.

Stuart indicated that if someone wants to try different components, it should work as well.

Carlos indicated that it is a very ambitious proposal, covering a large scope. Is there an overlap with other projects?

Margaret indicated that the goal of OPNFV is to try different combinations of components. If we have resources available to explore those components and combination, it should be fine. And the target of this OSCAR project is L3 VPN, which makes it unique. We should encourage to explore other combinations as well, such as containers. So if we have resources, fine. If we don't have resource, then it is a different story.

Dave Neary worried that this project introduced extra work, and not sure how much it will affect other projects. We should have projects that fill in the gaps that other project cannot cover.

Margaret suggested that TSC should set up the priority and focus. If we have resources, we need to be open.

Bryan commented that BGS project is to get off the ground. And this OSCAR project is to try different combination. We should avoid competition of different projects.

Parviz indicated that OSCAR already skips or has passed BGS / sandbox phase. It's a stable platform that has been tested, and ready to support testing various VNFs.

Bryan suggested that we should make sure that the components are plug-n-play in a compatible manner.

Parviz indicated that 3 new tasks are: docker/container, OVS v.s. vRouter, and different combination.

Margaret concluded that the biggest concern so far is staffing, and TSC should prioritize different projects, and set up focus.

Bryan emphasized that the project description reads to deliver an automatic installer and configuration, so it should support the choice of different components.

Chris said that we already have a CI project. OSCAR looks like to propose a different CI project.

Bryan explained that CI focuses on continuous build of code, including check-out repository, nightly build etc. But this project is to package and install different combinations of different components of choice.

Dave Lenrow indicated that if it is not tested by OPNFV test infrastructure, it means nothing

Margaret indicated that this OSCAR project tries different use cases and combination. It identifies a CI gap, i.e. OpenContrail component should be included in CI.

Chris concerned that this project proposed multiple categories, which covers multiple projects and seems to achieve the whole vision and goal of OPNFV.

Dave Lenrow also concerned that this project seems to use proprietary protocol instead of standard SDN control protocol such as OpenFlow. He suggested that we should start with an analysis of what needs to do as a starting point.

Chris indicated that BGS is the place to start with.

Uli concerned that he didn't understand how this project will work together with other project in OPNFV, because it seems OSCAR is done, and we just take it.

Margaret suggested and the group agreed that Bryan and Parviz will lead the effort of revising the OSCAR project proposal so that it fits better for OPNFV goal and scope, and the vision of OSCAR project. Cliff Young / Peter Lee, Dave Lenrow and Uli Kleber are also interested in joining the work of revising the project proposal. The group agreed to have another review meeting of the revised OSCAR project proposal at the same time tomorrow Friday Dec 12th so that it can be ready for formal TSC review and approval at TSC meeting Tuesday Dec 16th.

Bin will set up the call for reviewing the revised OSCAR proposal.

Because of time limitation, the rest of agenda will be discussed next week.

  * Discussion of Requirement project proposals
    * [[requirements_projects:Data Plane Acceleration]]
    * [[requirements_projects:OpenStack Based VNF Forwarding Graph]]
    * [[requirements_projects:Failure Prediction]]
  * Discussion of Collaborative Development project proposals
    * TBD
  * Discussion of other projects

Meeting adjourned
Wiki

User Tools

Site Tools

Page Tools
