What’s next for NFV in 2018?
In December last year, Virtual Strategy magazine published my predictions for network functions virtualization (NFV) in 2017.
That was fun, so let’s do it again! What does 2018 hold in store for NFV? Here’s my view:
1. The resurgence of proprietary management and network orchestration (MANO)
Open source management and network orchestration (MANO) projects such as OSM (Open Source MANO) and ONAP (Open Network Automation Platform) have gotten a lot of press and a lot of attention from operators. That attention includes dedication of resources to support the development of these projects. But the open source MANO projects are still immature and not ready for deployment. Real NFV services are moving forward with proprietary management solutions. These software packages are coming from traditional suppliers such as Ericsson, NEC, and Ciena. Even AT&T’s use of ONAP involves the support from the traditional supplier Amdocs. We may see a widespread move to deployment of open source MANO, but not in 2018.
2. The dominance of multi-vendor solutions
When operators invented NFV, one of their goals was to break the grip of closed single-vendor products. These operators wanted to move instead to multi-vendor solutions to increase competition and innovation. Some suppliers initially gave lip service to this direction but did not cooperate. This behavior was not going to change until service providers took a stand and required cooperation.
Initiatives such as Verizon’s universal customer premises equipment (uCPE) project signaled the end for closed single-vendor solutions. In 2018, we will see other operators follow Verizon’s lead and insist on multi-vendor implementations of NFV.
3. SD-WAN shifts from appliances to VNFs hosted on uCPE
One particular example of the benefits of the multi-vendor approach is SD-WAN. But it’s taken a while to get to those benefits. Service providers implemented most of the initial deployments of SD-WAN with vendor-supplied appliances. There were also a few bare-metal deployments of SD-WAN software.
Now service providers are moving to a cloud-native model, where they are deploying SD-WAN as a virtual network function (VNF) running on standard NFV infrastructure (NFVI). This NFVI is delivered in the form uCPE, which is based on Linux, KVM and OpenStack running on an open commercial off-the-shelf (COTS) server. This approach enables service providers to deliver a managed SD-WAN service, combine it with other functions such as firewalls, and do so in a manner consistent with their long-term strategy. In 2018, we will see a widespread move from appliances to uCPE as the delivery vehicle for SD-WAN.
4. Continued expansion of edge compute and micro-clouds
The initial focus of NFV was using existing data center resources for cost-effective deployments. This model works great in some cases, but there are many others where resources are needed out in the access network. We already discussed the example of uCPE for SD-WAN. Other examples include virtualized security, 5G infrastructure and IoT applications. There are also innovations such as “store in a box,” where the operator hosts the network and compute functions for a store or small office on a single low-cost server. All of these applications will drive increased deployment of compute elements in 2018.
Change is in store for NFV
How does that compare with your predictions for 2018? Whether I am right or wrong, it’s safe to say that 2018 will be a year of big changes for NFV.