DALLAS -- There was almost a meeting within the bigger Gen15 confab here this week, as a number of what we now call "SDOs" for standards development organizations were on hand, all generally present to tout the cooperative work they have been doing with the sponsoring MEF. Against that backdrop of agreement, however, they universally agreed there is still much to be done.
In fact, during a Wednesday afternoon panel, there was even some mild debate over what actually needs to happen next to push virtualization forward. Rich Schell, a product manager for Verizon Enterprise Solutions 's Secure Cloud Interconnect and co-chair of the OpenCloud Connect project's technical committee, said what he needs is a "killer app" that will attract people to his virtualized network and enable him to boost revenues. Thinh Nguyenphu, a Nokia NFV architect representing European Telecommunications Standards Institute (ETSI) 's NFV ISG, encouraged service providers to pick specific use cases for which they need standards now so SDOs can focus on those and get them accomplished. Meanwhile, multiple other panel members promoted the idea that a common information model is the best starting point, while also stressing the need for interfaces that enable carrier-to-carrier service delivery.
From the MEF 's viewpoint, its lifecycle services orchestration (LSO) effort has gone from zero to 60 miles an hour in terms of industry traction and was showcased here in a hackathon that attracted a wide range of coders, focused on how to provision CE 2.0 services from a software-defined networking environment. That work included developing APIs based on the LSO's Presto API framework and providing feedback for how that framework needs to evolve, and creating OpenDaylight SDN controller plugins that support LSO Presto APIs.
As Rami Yaron, co-chair of MEF's Global Marketing Committee, emphasized, the organization has been stressing cooperation with a wide variety of industry players on multiple fronts. So it's not for lack of trying that the industry hasn't settled on a single approach to orchestrating virtual and physical assets in the delivery of new services.
Most of these groups will be participating in the mid-January meeting in Louisville, Colo., which CableLabs is hosting to focus specifically on CIMs and how to rationalize the different SDO approaches. And that could be a watershed moment for industry cooperation in general. (See Can Telecom Unite on Info Models for NFV?)
There were a number of interesting additions to the general discussion that surfaced this week, however, underscoring the idea that, even as they cooperate, the SDOs are coming at the problem from their own vantage point:
- Everyone agrees more dynamic service delivery and particularly new services is key. Verizon's Schell made the strongest case, but even vendors on the exhibit floor were focused on highlighting how you use the technology to deliver new stuff. Often the focus of the services is based on the service providers' or vendors' piece of the bigger puzzle. CableLabs' James Kim, who was proudly sporting his LSO Hackathon t-shirt, nonetheless said his group is looking at virtualizing home networks and on plug-ins to SDN controllers that enable DOCSIS-based networks to invoke SDN assets or virtualized functions.
- One key "use case" is enabling connections between networks. MEF has always focused on interoperability and that was a key topic here. Pascal Menezes, vice-chair of the Open Networking Foundation 's NBI Group (northbound interface), talked a lot about east-west interfaces and the more horizontal connections required and multiple speakers pointed out the folly of enabling each network operator to create more dynamic services through virtualization if they can't interconnect to do end-to-end solutions across networks. And, as Menezes pointed out, the interfaces to do that are still undefined and lacking industry agreement.
- Industry-wide approaches are getting more practical. The LSO hackathon is one example of that, as are the proofs of concept and the TM Forum 's Catalyst projects, all of which attack real-world problems, noted Ken Dilbeck, VP of collaboration and R&D for the Forum. The good news is that network operators can be using some of the solutions developed through these efforts to get to market more quickly. The potential bad news may be creation of siloes of innovation that then have to be connected at some point.
- "Orchestration" as a term is getting harder to define. The word is being used to describe both broad approaches such as the LSO, and vendor-specific efforts that address only one set of gear or one ecosystem. As Nguyenphu pointed out in his Thursday morning presentation, the ETSI NFV ISG has moved away from its original terminology of MANO for management and network orchestration, to a working group called Interfaces and Architecture (IFA), which will actually issue its next set of specifications in February 2016. Others, including TM Forum, are looking to closed control loops within specific domains which are then abstracted to a higher level system that can be an orchestrator, or a choreographer. (See Major Change Afoot in Managing Virtualization.)
So while advances are being made and were showcased effectively this week at GEN15, there is still a lot to be done and still some confusion as to exactly what that work will resemble.
— Carol Wilson, Editor-at-Large, Light Reading