The rise of Open Platform for NFV: interview with Christopher Price

Dear Christopher, can you briefly explain what OPNFV is and the reasons of launching this community?

OPNFV is an open source project focused on accelerating NFV's evolution through an integrated, open platform derived from purely open source components.  The community was launched with the explicit intention of providing a reference platform for network operators to run large scale carrier applications in a virtualized or hybrid environment.

So, from your perspective, what are the major benefits of making OPNFV an open source project for vendors and for operators?

Being an open source project provides OPNFV with the ability to work in a completely transparent manner with our upstream open source communities. Source projects like OpenStack, Linux, OpenDaylight and others will be better able to understand what we are trying to achieve and how we are going about doing so as we work with the same methodology and in the same style as they do.
Our primary goal is to influence and develop features in our upstream communities; it is completely natural that we operate in the same manner as they do.
In addition as an open source project ourselves we create an environment where both operators and vendors are able to collaborate and the findings of those collaborations can be easily consumed by the industry and implemented quickly in products and services.

As we are at the OpenStack Summit after a successful OPNFV Track, can you elaborate about relationship with the ‘Upstream’ communities and the standard ecosystem? Do you think there is a new, more ‘hands on’ way to develop standards?

The signs are certainly there for change in the way we approach standardization.  It is not a new concept that a standards body uses software to validate proposals, the IETF has traditionally followed the concept of “loose consensus and running code” when approach new areas. 

The recent rise of open source as a development practice provides us with an opportunity to parallelize standardization and new development in such a way that there can be community consensus on both the definition and realization of features and functions in a way that makes the technology more adoptable.
I expect this will result in reduced lead-time from consensus to production, and will accelerate the evolution of our networks.

Release ARNO, the first OPNFV release, was released in June; what are the aims and what have been the challenges and the learning points?

With Arno we decided early on that as a community we needed to understand how we would work together than what we needed as far as our infrastructure and methods were concerned, we started with a blank slate and essentially gave ourselves six months to perform a first release.  The TSC very quickly determined that the Arno release would need to focus on a few achievable goals, we elected to establish the OPNFV projects CI/CD pipeline, our global lab infrastructures, and building a running platform for developmental purposes to launch our initiative.
While this may sound like a reasonable scope there have been real challenges even in establishing the foundation for OPNFV in such a short timeframe.  Without going into details I can say confidently that we now have a very good idea of what will be required moving forward and how we need to be organized for success.

There is a lot of discussion about DevOps, Continuous Integration and Continuous Delivery. OPNFV is striving to make them real; can you give us some examples of adoption of those methods within the OPNFV projects, e.g. BGS and Octopus?

DevOps is as much a way of approaching tasks, as it is in the way you perform them.  In our Arno release project we started with a blank slate and consciously relied on and promoted automation, cooperation & communication as the key tools for our development teams.  Our activities for Arno start with needed code, to integration, to deployment and validation.  Approaching those challenges with a singular purpose and goal, as a collaborative community, was the only way we were going to be able to achieve success.
This BGS, octopus and functest teams each worked toward a common goal and with full visibility into the actions and intentions of each other team.  A challenge for one team was a challenge for all teams, was seen as such, and was addressed collaboratively in true DevOps fashion.  We are fortunate in that meritocratic open source development not only promotes, but also requires such collaboration to be effective.
DevOps might not come naturally to the telecommunications industry, but the OPNFV team has provided a testament on how we can adapt and accelerate our activities using state of the art tools and practices.

Regarding the adoption of OPNFV and Arno Release, do you have any early signs? What they are and what /where are the main labs in which the ARNO release is being built and tested?

OPNFV adoption is one of those key items we of course want to see as an outcome of our work.
Our Arno release, while developmental and not yet providing the stability and scale we intend, provides a deployable well defined reference platform that is available for everyone to begin to work with and collaborate on.  We fully acknowledge we have a lot of work remaining until we have developed a platform suitable for VNF deployments and commercial adoption.  Arno provides us with that first stepping stone, setting a direction, demonstrating progress and establishing a baseline for us to work across the industry raising the bar step by step, release by release.

What advice would you give to someone just getting started in OPNFV?

OPNFV can be a little daunting when you are trying to find your way in the community.  Our scope is very broad and it is easy to be swamped by the breadth of activity across the project.  I would suggest for anyone trying to get started to reach out to a TSC member or community leader and start with a simple conversation.
Explain why you are interested in the community and what you might want to do, our leaders will help you find like-minded community members who will be more than happy to help get you started.

As final consideration, what is your vision of OPNFV two years since now, where do you think this community is headed ?

I expect in two years OPNFV will be providing a fully functional reference platform for the industry that is not only understood by hardware and platform vendors but addresses the needs of the application development community and operators.
OPNFV will always be a mid-stream activity, it’s value is in it’s ability to provide a common ground across the industry for all members to share and align on key items necessary to drive change across our global networks.
We see the sign already today that this is the right place to come together to forge common capabilities, over the coming 24 months these features will be realized and further iterations of the platform will improve it’s capabilities, usability and flexibility for every-one involved.

 

Torna all'articolo


Christopher Price (Ericsson)

leads open source industry collaboration for Ericsson in the areas of NFV, Cloud & SDN from the CTO’s office in Sweden.  He is the chairman of the OPNFV Technical Steering Committee and is an active member of the OpenDaylight Committee. Chris’ experiences include leading Ericssons' IP&Broadband network architecture and standardization teams with a rich history in development of systems and technology in the areas of network management, policy control and user service management, user session control plane solutions, and DPI technologies