Torbz - Fotolia
Every living thing is part of a larger ecosystem in nature. The same holds true in technology. Making changes to how networks are built affects everything, from service ordering to problem management. All these changes must be considered when trying to generate a business case for technology shifts, such as software-defined networking and network functions virtualization.
Standards groups and industry forums tend to focus on the technology target, not the ecosystem, which can quickly dissipate the technology's benefits.
In terms of the benefits of SDN and NFV, the promise of Capex reduction initially justified the implementation of both, as it does with many developing technologies. But operators and other insiders now doubt Capex alone can really drive a rapid and complete adoption of either technology. The question becomes how to harness other benefits of the technologies.
Fortunately, the following three technologies can help build value to increase the benefits of SDN and NFV.
1. Orchestration and service modeling
The first of the technologies that can help build the value of SDN and NFV is service modeling and top-to-bottom orchestration. The question becomes how to harness other benefits of the technologies -- and a development in NFV provides the beginnings of a solution.
The core of the NFV functional model -- published by the ETSI NFV Industry Specification Group -- is management and orchestration. This particular NFV feature is designed to take a model of a virtual function and turn it into a running equivalent of a traditional hardware device, like a router or firewall. The ETSI NFV ISG divides orchestration into two layers: service orchestration and resource orchestration. This extension lets operators or vendors build functional models of services that can then be orchestrated to not only deploy virtual network functions, but also control legacy network components.
The key to orchestration is the model, and SDN and NFV modeling converges on the concept of intent models -- abstract descriptions of a technology system's external features. A virtual network access device might consist of a half dozen chained virtual functions, but its intent model would show only its features, not its structures. Consequently, anything that can realize the features can be deployed, facilitating the evolution to SDN and NFV technologies from legacy network devices.
If services are modeled from the top down, then orchestration can replace manual provisioning to significantly reduce operations costs. Service changes and recovery from failures can be implemented more quickly, and new services can be introduced as soon as the opportunity or competitive risk is recognized. Modeling and orchestration is the most important technology for SDN and NFV.
2. SD-WAN functionality
Software-defined WAN (SD-WAN) is another technology that benefits SDN and NFV. SD-WAN technology builds an overlay network by defining a protocol above the Data Link Layer (Layer 2) or the Network Layer (Layer 3) that is carried in tunnels by the physical network. This overlay network can be laid across any lower-layer technology that can support a tunnel or virtual wire. Connectivity emerges from edge devices that have knowledge of the virtual network tunnel topology. It is even possible to tunnel over the public internet.
This SD-WAN approach is often used to separate users and tenants in public cloud services. Each SD-WAN is independent and isolated. If the underlay network technology offers classes of service, SD-WAN can deliver the same quality of service and service-level agreements as traditional MPLS virtual private networks or Layer 2 virtual LANs. SDN can route SD-WANs and also build underlay network tunnels. SDN can be used to build services with NFV-hosted functions. SD-WANs are easy to control using service modeling and orchestration and extend the functional model of service into the real world of devices.
3. Open source promotes collaboration
Open source also increases the benefits of SDN and NFV. Both SDN and NFV suffer from two problems that have proven difficult or impossible to overcome in the past. The first is a bottom-up approach taken by traditional standards organizations when software-based technologies should be designed from the top down -- moving from benefits to implementation. The second problem resides in equipment vendors' resistance to changing technology -- changes that threaten their near-term profits.
An open source project builds software directly. Even if the project launches from bottom-up specifications, the fact anyone can contribute features to open source code means it can shift direction and accommodate a top-down, benefit-driven approach. Additionally, multiple open source projects are launched to support a specific mission -- as is the case with both SDN and NFV. Each of these projects normally takes a different approach, and the market will pick the one that can deliver the most benefits the fastest.
Open source is based on open source code that prevents any vendor from owning the approach and limiting its pace of adoption. If vendors adopt their own proprietary approach, the availability of a parallel open source service limits how much the vendors can lock in customers or charge for their software.
Current benefits of SDN and NFV
A combination of these three technologies could advance both SDN and NFV considerably, perhaps even decisively. But what are the chances this will happen, and when?
Service modeling and top-to-bottom orchestration are already emerging in some vendor offerings and in standards group discussions. Network operators in the U.S. and Europe have published their own SDN and NFV architectures that include modeling and orchestration capabilities. AT&T began making its entire architecture -- Central Office Re-architected as a Datacenter (CORD) -- open source. This is a critical step and a prerequisite for progress elsewhere.
A number of proprietary SDN vendors already use SD-WAN overlay network concepts. These vendors now link their offerings to both cloud computing and NFV. But these initiatives haven't been applied directly to SDN, NFV or cloud in an orderly way -- perhaps because the service modeling and orchestration evolution isn't quite complete. Advancing SD-WAN technology may, thus, be the current target for those hoping to move SDN and NFV forward quickly.
Open source is the darling of the industry. Clearly, mainstream SDN and NFV open source projects have advanced, but if AT&T found it necessary to first create an architecture of its own and then open it for broader use, there's clearly something missing. AT&T's CORD, hopefully, will crystalize what we need to see from open source.
There's no such thing as an independent technology in an ecosystem like networking. Both SDN and NFV will have to identify symbiotic activities and harness them to get the most from their own efforts, and some of that is already starting to happen. All three of these critical technologies -- service modeling and top-to-bottom orchestration, open source and SD-WAN -- will likely make positive contributions to the network of the future.
SD-WAN and NFV benefits in software-defined networking
Building an open source business case
How SDN and NFV technologies are changing IP