Onboarding VNFs has long been considered a major challenge for network architects looking to create best-of-breed multivendor environments in their transformed virtualized networks, with network functions and use cases spanning the gamut from ‘carrier-grade’ infrastructure to enterprise IT. In the telecommunications sector alone, this year saw the emergence of a range of open standards and open source initiatives to tackle this challenge, notably in the TMF Forum Zoom and ‘well-enabled VNF package’ initiatives as well as in the new open source communities of OPEN-O and ETSI Open Source MANO (OSM).
As a panel moderator working within a fixed hour-long time window, six expert panel participants and a large engaged audience bursting at the seams to ask questions on this complex issue, the passion in the discussion was only matched by the intensity of the klieg lights illuminating the stage. In such a situation, as a moderator I find it important to conclude a panel session with concrete bottom lines take-aways for the audience. With that in mind, here is a summary of what the panel of experts advised the audience when considering their VNF onboarding and MANO standardization strategy:
- Lakshmi Sharma (Brocade). Lakshmi focused on the practical steps to solve the broader virtualized network function deployment challenge. She advised to start by defining workflows (no matter whether you are in a Communications Service Provider (Comms SP) or enterprise IT network), automating the workflows and offering them as a service (‘automation as a service’), and then building the platform so that it can be automated as well (‘automation as an infrastructure’).
- Guy Machlev (Nokia). Guy highlighted the need to first define the abstraction, then the Application Program Interface (API), and finally how to automate the VNF on-boarding and deployment process.
- Amir Levy (Gigaspaces). Amir expanded on Guy’s API point by outlining the work that Gigaspaces is doing on their Software Development Kit (SDK) and tooling to allow VNF vendors to get their VNFs cloud-ready. In addition, Amir highlighted the leadership role that open source consortia, such as OPNFV, OPEN-O, OSM and ECOMP, play in creating an orchestration layer to provide the desired ‘single pane of glass’ capabilities.
- Sanjay Aiyagari (Red Hat) agreed with Amir on the importance of open source communities and open interfaces. He then suggested to think about VNF on-boarding from the perspective of carrier service level agreements rather than the physical device. Lastly, he argued that the visibility into the network (i.e. the integration of telemetry) is a critical consideration when developing a comprehensive MANO and VNF on-boarding strategy.
- Dave Votava (F5 Networks) stressed that agility is needed to work in the various environments and believes that DevOps type of approaches should be considered as part of the VNF strategy. Similar to the Brocade position, he then highlighted the importance of thinking through the consumption model for the network services, no matter whether they are optimization, scaling or security related services, since it is impacted by programmability and orchestration considerations.
- Les Stuart (HPE) concluded the session by speaking to the need to consider both the NFV stack as well as the physical network for end-to-end service orchestration. Providing the example of a virtual customer premises equipment (vCPE) network, he emphasized that many customers will want to leverage the existing legacy networking equipment (managed by Element Management System; Network Management System; Operations Support System) that might already exist on the customer premises. Les agreed with Sanjay on the importance of telemetry for service level monitoring, especially closed loop management and automation. In his final statement, Les then reminded the audience that NFV is a transformation that we have just started.
Clearly, VNF onboarding and MANO across different network environments is a complex and multi-faceted area. I was encouraged by the common threads that started to emerge in the conversation, however. In particular, the focus on implementing and orchestrating end-to-end services in an agile DevOps fashion, the importance of open source consortia to drive interoperability, as well as the need to closely tie VNF orchestration strategy to telemetry are three critical messages for anyone considering the next stages in their transformed NFV network. Continue to join us here in this blog channel as we are going to continue exploring these topics further in 2016.