|
Key Takeaways
|
- Leverage software-defined networking (SDN) concepts to boost flexibility and scalability.
- Implement network function virtualization (NFV) technologies to reduce hardware dependency, decreasing costs and improving scalability.
- Employ SDN and NFV to create more dynamic environments and enable agile networking.
|
SDN and NFV are acronyms you hear frequently in discussions of modern networking. In fact, they appear so commonly that they can be easy to confuse or conflate with one another.
But that would be a mistake. SDN and NFV are related terms, but they are also distinct. You can use SDN without using NFV, and the benefits of NFV are not the same as the benefits of SDN in general.
Keep reading for a breakdown of what SDN and NFV have to do with each other, and what to use when.
What is SDN?
Software-defined networking, or SDN, is an approach to network design and operations that relies on software-based controls. With SDN, you can define and manage network virtual resources using software. Those resources are distinct from the underlying physical network.
For example, imagine that your company operates multiple sites. With traditional networking, each site would have to have its own network because there would be no way to create a single network spanning multiple geographic locations. But using SDN, you could create a software-defined wide area network (SD-WAN) that spans multiple offices. Doing so would in turn make it easier for each site to access the same network resources, as well as to manage networking in a centralized way.
In the context of SDN, the underlying physical network is usually referred to as the underlay network. Virtual networks that you build on top of physical networks are overlay networks.
SDN has been widely adopted over the past decade mainly because it provides much more flexibility and scalability than organizations can achieve using physical networks alone. Put simply, SDN gives you many more options in the way you design and manage your network.
What is NFV?
Network functions virtualization, or NFV, is the use of virtualization to create and manage network equipment or services.
For example, rather than relying on a physical device to serve as a load balancer for your network, you could implement a virtual load balancer using NFV. Businesses can likewise adopt NFV to virtualize firewalls, border controllers, and almost any other network service or function that would have traditionally been implemented using physical devices. When you virtualize a network function using NFV, you turn it into a virtual network function, or VNF.
Like SDN, the main benefit of NFV is that it provides more flexibility and options. NFV frees organizations from being bound by physical devices in order to implement the network functions or controls they need. It's easier, faster, and (in most cases) less costly to deploy a virtual network function or service than it is to set up a physical device to implement that service.
SDN vs. NFV
The fact that SDN and NFV both make it possible to virtualize parts of your network doesn't mean they do the same thing.
The main difference between SDN and NFV is that SDN virtualizes the network itself, whereas NFV virtualizes network services or functions. In other words, SDN allows you to implement virtual network architectures, while NFV replaces physical network equipment and services with virtual alternatives.
So, you could use SDN to abstract your network from the underlying hardware in order to streamline and automate network management. But you can't use SDN to turn a physical network device – such as a load balancer – into a virtual device. You'd need NFV for that.
SDN and NFV: better together
It's possible to use SDN and NFV independently of each other. You could have a software-defined network that includes a physical load balancer, for example, and you could deploy a virtual load balancer on a traditional network that is not software-defined.
However, SDN and NFV go hand-in-hand because they both allow you to create more dynamic environments and enable agile networking. By using SDN and NFV together, you get maximum flexibility and, in general, a lower total operating cost for your network.
The challenges of SDN and NFV
SDN and NFV are also similar in that they create the same basic types of challenges. In particular, they both increase the complexity of network management and monitoring operations.
SDN does this because it adds overlay networks into your network architecture, increasing the complexity of the network topology. When a problem occurs on a software-defined network, you have to figure out whether the issue originates from your overlay network or the underlay.
For its part, NFV makes network management more complex because it tends to increase the number of logical components you have to track and monitor. Because deploying VNFs is so easy, you're likely to end up with more functions or services on your network, and those functions and services are likely to change more quickly.
In addition, in many cases VNFs depend on containers or virtual machines to host them, so you need to monitor your VNF host infrastructure in addition to the VNFs themselves. You also need to track dependency mappings so that you know which VNFs depend on which containers or virtual machines, and how the failure of one part of your hosting infrastructure could impact network functions.
The added complexity of SDN and NFV is usually worth the cost, since it leads to greater network agility. But it's still a challenge to be aware of if you decide to take advantage of virtualization.
Conclusion: preparing for SDN and NFV
Indeed, SDN and NFV bring so much value to modern networking that it's difficult to imagine running an efficient network without them. But it's also difficult to imagine managing a modern network efficiently without monitoring tools that are capable of handling the special challenges posed by both SDN and NFV. To take full advantage of modern networking techniques and technologies, you need monitoring tools that can support multi-layer visibility, service chain views, and dynamic relationship mapping.
