2019 is coming to a close and 2020 is right around the corner! Now is a time to reflect on what has happened in the past year and what will happen in the new year. This is part one of a two-part blog post highlighting what is driving the trends and what I expect to see. In each part are a couple of predictions of what I expect to see happen this coming year.
The big prediction that we have been hearing this year is that NFV and SDN are dead. In essence, these are now just old buzzwords, but “are they really dead?” is the question. All of this came about after research firm Gartner claimed that they are dead. What has been most interesting to me has been seeing all the reactions to this. It was clearly a sensational statement made to get our attention, but I think this has been very good at creating good thoughtful conversations that wouldn’t have occurred otherwise. What I see is that SDN and NFV are evolving into what’s next. The new big buzzword may be Intent-Based Networking (IBN).
The death of NFV and SDN really isn’t that surprising. Many have stated that Cloud has killed NFV and/or SDN. This isn’t necessarily wrong. However, instead of focusing on why or if they are dead, I see it as a change in how network services are deployed and delivered. In a cloud-native approach, micro-services are deployed by using containers and things like Docker and Kubernetes. In this case, the architecture changes depending on what it runs on, but the management, configuring, and monitoring are all still pretty much the same. How the network is managed and what is being managed will evolve, but at the end of the day, the ability to configure and monitor it is still required.
The best way to do all of this is through programmability with NETCONF and YANG. By moving from physical infrastructure to NFV and SDN there are more things to manage and it quickly becomes much more complex in its application to the network. Imagine how much more the complexity has grown moving from these to the cloud. Programmability enabled automation is still the key foundation on which to build for managing this growing complexity.
2020 will see programmability and automation becoming more and more important to make our network elements more connected and integrated across the network. At the end of the day, the network must be programmable down to the lowest level. No matter what you call it, there is still a strong need for programmability in order to meet the future needs head-on. With the right foundation in place, you can build anything.
The second prediction regards how I am seeing that programmability is starting to consolidate and standardize above the network elements. I am seeing more and more conversations around service models vs. element models. As a matter of fact, there is IETF work happening right now on models with service assurance as the focus.
Programmability at the device level was at the beginning of what Tail-f Systems pioneered. Our solutions eventually grew to include Cisco NSO which is now an industry-leading orchestrator. We developed this in the very early days of modern network programmability and no one had been thinking of using YANG for modeling services. There was a mismatch in this industry about how to bridge the gap between devices and services. While it sounds simple, this was a very hard question the industry was trying to deal with for almost 10 years ago. Then with NSO, we showed how YANG modeling could grow up the stack and not be focused just on the device. By applying YANG modeling to both services and devices, we could turn network management into a simple computer science problem of mapping YANG service models to YANG device models.
The world is recognizing that what we did at this point was critical in delivering orchestration and automation at the service level. The next step was creating a standard with regards to how these models function. This is what the IETF did with RFC 8199 which gives definitions of abstraction levels for YANG models. Given this evolution, it is interesting to see all of this rollover to service assurance work. I read a great article by Benoit Claise, IETF Area Director for Operations and Management, on how he sees this evolving. He sees service assurance as a critical part of Intent-Based Networking (IBN) with service assurance being modeled in YANG. We expect to see things like this implemented more in 2020 as large enterprise and service providers find new innovative ways to tell the network how to be configured and to verify that the network is operating as intended. Once organizations get to this point, Artificial Intelligence (AI) will take the intent and make it real in the network and ensure that it stays real. Again, this evolution will be built on the foundation of network programmability.
We have looked at the death of NFV and SDN and the great integration of programmability across the network. Stay tuned for the second part of this blog as we focus on two more trends: the death of the CLI and network admins becoming network programmers.