By Jason Hunt, LFN Technical Steering Committee Chair
(With contributions from Abhijit Kumbhare, Al Morton, Catherine Lefevre, Gergely Csatari, and Prabhjot Sethi)
The LF Networking community is gearing up for its next round of LFN Technical Meetings, taking place virtually February 1-4, 2021. The agenda is still underway, but we expect another fantastic line-up of collaborative sessions designed around building more open, interoperable networking stacks.
More specifically, sessions will focus on the hows and whys of progressing LFN project releases as well as project architecture, direction, and integration points. Normally our LFN technical events are held in-person, giving community members the opportunity to meet informally and share social time together. However, under the current circumstances, this event will be held virtually, via a combination of online tools to create a productive and interactive virtual meeting environment. The event is free of cost and registration is required.
For more details –including how to register — please visit https://wiki.lfnetworking.org/pages/viewpage.action?pageId=46106257
We’d also like to extend a vote of thanks to all the attendees of our recent Linux Foundation Networking (LFN) Technical Event, held October 13-15th. The purpose of these events is for our LFN project communities to go deep on the technical aspects of their projects while also fostering collaboration across the LFN umbrella. The event planning team did a great job balancing all the sessions to maximize the return on our time spent together.
While nothing can replace that interaction, the virtual events do allow us to broaden our participation to those who wouldn’t normally be able to travel due to budget or time constraints. We’re happy to say that 362 people participated in 75 sessions during our October event. This high level of participation bodes well for the future of the LFN projects and our impact in the industry.
Read below for details about outcomes from each of our four track sessions.
Key Takeaways from Project Tracks:
CNTT (with OPNFV)
The CNTT and OPNFV communities used the LFN Technical Event to discuss the technical aspects of the Elbrus release and to discuss the merger of OPNFV and CNTT communities.
The Reference Model (RM) team hit topics including:
- Adding network and hardware management to the model with the focus on how workload affects of these factors
- alignment of workload and infrastructure security requirements to ONAP,
- abstraction and orchestration of accelerators with alignment with O-RAN, storage requirements, modeling the infrastructure for hybrid clouds and building special use cases to the model like load balancing, service discovery or service function chaining.
The Reference Architecture for Kubernetes (RA2) team conducted a survey on the usage of Kubernetes in telecommunications. The results of the survey were discussed in one of the sessions. It is apparent that all network operators who answered the survey are using Kubernetes, and 50% usesare using Kubernetes to run workloads in production. This shows that containerization and cloud native is not the future anymore, but it is the technology stack of our present time.
RA2 discussed the problem of profile granularity, which results in optimal usage of resources. The problem is that the current profiles defined in the RM are”all or nothing” for cloud native workloads and there would be a need to express and support more fine grained requirements within the infrastructure. This needs to be discussed with the Reference Model team.
The Reference Implementation for Kubernetes (RI) team reported the work status of the current implementation of the OPNFV Kuberef project, and planned how to close any gaps. In the OvP session it was discussed that the release of the first Cloud native OvP badging will be synched to the release of CNTT RI2.
The Reference Conformance for Kubernetes (RC2) team focussed on CNF requirements. As the mandate of CNTT is to make the integration of cloud infrastructures and VNF/CNF-s by defining the interface between these two only cloud infrastructure-related CNF requirements should be defined in CNTT.
OPNFV met jointly with CNTT to work out more details of their merger, and reviewed the results of a Kubernetes user survey that will guide OPNFV development projects as they continue their pivot to cloud native work/K8s.
Other highlights included:
- During the event, OPNFV’s TSC approved milestone completion ahead of releasing OPNFV JERMA on December 15 (the TSC always meets during technical events).
- Review of Edge Workstreams identified profiles need to be addressed in the reference architectures and model.
- OVP 2.0 discussed CNF Workload requirements — requirements for the infrastructure make sense, and there should be a test of the infrastructure to ensure that the workload requirements are met.
- Sessions investigated CNTT Telemetry Requirements with AIRSHIP and Barometer projects, and reviewed the relatively new Kuberef project, which relies on BMRA (new) installer for now, helped by testing the new Gitlab CI/CD environment from a green-field point of view.
The OpenDaylight community had sessions in the following areas:
- ODL Deployment & Use Cases.
- Kernel Projects: Aluminum release post-mortem and look-ahead for the Silicon release.
- Platform Projects and Simultaneous Release Mechanics
- A deep dive on the OpenDaylight Micro Project.
Of particular interest was the talk around ODL Deployment and Use Cases (including potential new use cases). While traditional use cases like NetConf & OpenFlow/OVSDB were discussed, there were questions around using ODL for ACTN controller for Hierarchical SDN as well as in the optical space. It was noted that ODL has been used in these spaces (Transport PCE use case for optical networking as well as a PoC for ACTN).
We invite any interested parties to join the ODL community! We are looking for more collaborators to help with key areas such as release pipeline (integration test & release engineering) as well as other key projects. Visit https://www.opendaylight.org/ to get involved.
The ONAP community used the October Technical Event to focus on three themes:
- Planning for the upcoming Honolulu release
- Long term requirements and best practices
- A demo track
Sessions in the Honolulu planning theme allowed community members to learn and provide feedback about upcoming plans for various use cases. Key Highlights include:
- a session on 5G network slicing, where a lot of progress has been made, with good discussion of priorities for new features for the next release.
- sessions dedicated to new and exciting features coming in the Honolulu release (stay tuned!).
- lively discussion around the shift of ONAP to address the needs of modern cloud native networks.
- Community members who collaborate with standard organizations like ETSI, 3GPP and TMForum shared updates from these organizations and led a discussion on alignment with ONAP. Similar updates were provided about alignment with external open-source projects like the O-RAN-SC.
On the long-term requirements track we had a chance to raise awareness across our community to topics like Control Loop, Security and Documentation. Other sessions in this track covered plans to modernize the overall architecture by adding components such as a service mesh. The developer forum provided an excellent opportunity for community members to discuss these topics that often don’t get the attention they deserve due to urgency of some shorter-term work.
Sessions in the demo track were designed to educate folks about the amazing achievements we have made recently. Attendees of these demo sessions got a chance for a close look at some exciting new features and use cases: 5G core orchestration, O-RAN cell handover, multi-tenancy and optimization framework are just a few of the demos shown. Community members were able to interact with the demo presenters, raise questions and suggest ideas for improvements.
Overall, the community had a chance to go beyond the day-to-day tasks and spend time planning the future of ONAP, while celebrating our recent success stories.
Tungsten Fabric sessions at LFN virtual technical meeting were centered around following agenda:
- Transparency of governance and release management with broader community
- Future Roadmap and Outlook for Tungsten Fabric as a project
- Showcase of some active projects which developer and user community can benefit from
Keeping the TSC elections scheduled February 2021 in view, we had ensured focus on fixing and enhancing governance documentation, inviting new participation in TSC. Additionally, setting up a release plan and process was a long standing Tungsten Fabric TSC goal, which was achieved recently and corresponding details were shared with the community. Tungsten Fabric is scheduled to do a 6 month release cadence with next release upcoming by end of November. Appropriate Project operations, process and milestones details were discussed and updated as part of documentation.
Roadmap discussion allowed the community to set up the future direction of the project with respect to traction for both OpenStack as well as Kubernetes based deployments addressing use-cases, along with setting expectations around the possible improvements in the core SDN functionality. Supporting Cloud Native use-cases was identified as the next step and detailed discussion was done on various functionality, features and use-cases to be addressed in this domain
Additionally these discussions helped identifying a series of technical documentation that are still residing in OpenContrail google drive and needs to be imported as part of tungsten fabric documentation