What is Network Virtualization?

What is network virtualization?

Network virtualization is the process of decoupling network services and functions from the underlying hardware. These network functions include internet protocol (IP) routing, packet switching and filtering, load balancing, and virtual private networks, and are combined into a single pool of resources in virtual networks.

Where traditional IT infrastructure served up networking resources on physical servers, routers, and switches, virtualized networks create a logical overlay that allows you to treat most network functions separately from the hardware. This means you can create, deploy, provision, and manage virtual networks in software, only relying on the physical infrastructure to handle IP packet forwarding.

With network virtualization, you can consolidate a group of multiple physical networks into a single virtual network, separate a single network into multiple virtual segments, or connect virtual machines (VMs) across distinct domains. Virtual networks can be isolated from each other, even though they all share the same underlying physical hardware.

Similar to how server virtualization enables the creation of multiple virtual machines that reside on a single physical server, network virtualization allows you to have multiple virtual networks that run atop a single physical network infrastructure.

Network virtualization represents a significant departure from traditional IT infrastructure because it frees you from physical hardware limitations and the challenges of physically networking VMs. It provides a great degree of flexibility because you can group or separate virtual networks as needed—setting specific networks aside for designated workloads, for instance—or organize and connect VMs however you want.

Because there’s no need to configure the physical hardware to run or create virtual networks, scaling up or down can be fast and easy.

How does network virtualization works?

In a virtual network, specific network functions and resources are sent out to the virtual layer, which is managed by a hypervisor that serves as a virtual network adapter. The virtual layer includes all the nodes and links that comprise the virtual networks, and the hypervisor assigns resources to workloads as needed. Each virtual network is isolated from the others and can have policies that are different from the other networks. 

One of the things that organizations like most about network virtualization is that once you’ve configured the appropriate network and security policies for a virtually networked application, you can move that application to another host (server) and those network and security configurations go with it. Even better, if you need to create new workloads to scale that application, those new workloads automatically get the same policies you’ve already configured. This helps keep policies consistent and saves IT time and effort.  

Hosts and workloads such as VMs can communicate with other hosts and workloads on separate virtual networks, using specific host protocols and virtual switches and routers. One advantage of this workload-to-workload communication is that it doesn’t travel over the physical network, so latency is reduced.

Why is network virtualization important?

Network virtualization, like other types of virtualization in the enterprise, is an extremely common practice among modern organizations. It allows IT to optimize utilization of physical infrastructure, reduces the management burden, and enables significant flexibility compared to traditional IT infrastructure. Virtual networks are more dynamic, efficient, and scalable.

The efficiency, flexibility, and agility provided by network virtualization are increasingly becoming must-haves for organizations that want to stay competitive, meet evolving customer needs and market trends, and deliver on the promised benefits of the cloud. Virtualization allows you to provision resources or update applications in just minutes, scale up or down as needed, and speed up overall time to market.

Types of virtual networks

There are two main types of network virtualization: 

External network virtualization  

This type allows you to virtualize physical networks that are on the same local area network (LAN), which could span multiple physical servers. Using virtual switches and adapters, external virtual networks can also virtualize multiple physical LANs into a single virtual LAN (VLAN). Organizations use this type of network virtualization to make management more efficient—because IT can manage and configure multiple systems connected to the VLAN instead of working on each machine individually. 

Internal network virtualization 

This type is confined to one network server. It can help improve performance because it simulates a physical network through the use of distinct software containers, relevant code, and all the various dependencies needed by an application to perform consistently across the entire ecosystem. All the VMs that reside on that server can communicate with each other without having to use an external network. 

Another set of network virtualization categories is defined by where they’re used in the network—the data center, wide area network (WAN), or LAN. Software-defined networking (SDN) was the main driver of network virtualization in the data center, thanks to SDN’s more centralized control and automated resource management. And software-defined WAN (SD-WAN) and LAN (SD-LAN) made it possible for organizations to virtualize WANs and LANs to improve efficiency and streamline processes. Virtual WANs enable IT to optimize bandwidth utilization for critical applications and other workloads. Organizations often use VLANs to manage and control network traffic more effectively and to create separate virtual networks.

Network virtualization and cloud computing

Virtualization and cloud computing are technologies that complement each other. Many organizations use virtualization to reduce management complexity and pool cloud-based network resources. They also use network virtualization in the cloud to build multiple virtual networks across one physical server or other infrastructure.

Virtualizing networks for cloud computing environments enables organizations to:

• Optimize resources more efficiently – With multiple virtual networks on a single hardware infrastructure ensures that bandwidth and resource utilization will be optimized, which can help improve network performance and minimize traffic congestion.
• Enhance security through isolation – Each network can have distinct security policies and isolation that ensure a breach won’t affect any other virtual network on the same infrastructure.
• Increase flexibility and scalability – You can create virtual networks that meet unique workload needs and speed up provisioning with dynamic automation and scaling on demand.
• Simplify cloud management – Because virtual networks provide centralized control, they help simplify tasks like configuration, monitoring, and troubleshooting in the cloud.
• Improve disaster recovery and ensure business continuity – Virtual networks can be replicated or moved from host to host as needed or even to on-premises data centers. With this redundancy, disaster recovery and business continuity are improved—even if the worst happens.

One interesting way to benefit from network virtualization in the cloud is to create virtual private clouds. If you have workloads on the public cloud, you are likely sharing physical hardware with other customers and users. By virtualizing your cloud network, you can enjoy a dedicated, isolated pool of network resources that belong only to you, with its own security policies, routing tables, subnets, and IP addresses. This allows you to control your network environments and define network policies and topologies as you wish.

Benefits of network virtualization

• Simplify network management – Virtual networks reduce the management burden in regard to physical  hardware, since it no longer needs to be configured. With virtualization technology, many management tasks, both in the data center and in the cloud, are automated, which helps save IT times and reduce management costs.
• Increase network flexibility – Virtualization allows you to move resources and workloads where you need them, whenever you want, regardless of the physical hardware topology. Scaling is fast and easy, too, so your organization can respond more efficiently to fluctuations in market trends or customer preferences.
• Support remote and hybrid work models – With a shared pool of resources across virtualized networks, employees can stay productive wherever work takes them.
• Boost security across networks – Isolation of virtual networks helps improve security because while an attacker may get into one network, it’s separated from the others, which are not affected. Isolation can also keep sensitive data protected from the rest of the infrastructure, whether on-premises or in the cloud.
• Speed up time to value – Thanks to automation and reduced management burden, network provisioning time can be reduced from days or week to just minutes. That allows great agility and efficiency, which saves time and money.
• Reduce costs of hardware – With the ability to create multiple virtual networks on a single server you can reduce your physical footprint, which saves money. Less hardware could also mean a reduced need for physical space and reduced need for heating and cooling—which also saves money in the long run.

Challenges of network virtualization

Network virtualization comes with many benefits, but it can also introduce a few challenges as well. These could include: 

• Virtual network sprawl – Because virtual networks are so easy to create and provision on demand, it can sometimes lead to a large number of virtual networks that are no longer needed (because the project is complete, etc.). These networks could be out there consuming resources, costing you money—especially in cloud environments where you only pay for what you use—and adding complexity to your infrastructure.
• Considerations of changes in network architecture – When moving workloads and other data to virtual networks, it might change how the workload consumes resources or initially affect factors such as resilience and security. Those issues can be worked out, but there might be a steep learning curve at first.
• Lack of collaboration between teams – Virtualization merges some disciplines that were traditionally siloed, such as networking and security, for instance. It might be difficult at first to keep teams collaborating to ensure the entire infrastructure is running as it should.
• Potential skills gap – Because network virtualization is a whole different approach to computing, it might be challenging to get IT employees up to speed on how to manage, configure, and maintain virtual networks.
• Shifting visibility into networks – The layer of abstraction between virtual networks and physical infrastructure could mean that the monitoring tools you’re used to no longer give you the visibility you need into networks. However, there are many tools available that are optimized for virtualized networks and other infrastructure.
• Added complexity due to automation and AI – Automation and artificial intelligence (AI) can be game changers when it comes to virtualization but if your team isn’t familiar with how to deploy and manage them it can add complexity.

Examples of virtual networks

The following are some common examples of network virtualization:

• VLAN – A segment of a physical LAN that is virtualized to combine network nodes and/or devices on multiple LANs into a single virtual network with a shared pool of networking resources. VLANs are often used to boost performance on high-traffic networks and make provisioning and updating networks fast and easy.
• Network overlays – There are many types of network overlays—VLANs are a simple one—but the term typically refers to more complex virtual layers that sit atop physical hardware infrastructure. Overlays allow IT admins to configure, define, and manage the flow of data traffic without consideration of the underlying physical hardware.
• Network functions virtualization (NFV) – This strategy, often used by mobile telecom carriers, entails the removal of specific network functions, such as firewalling, routing, and load balancing, from dedicated physical hardware and placing them on virtual servers or (sometimes) off-the-shelf commodity hardware. It allows carriers to take advantage of specific features without having to purchase specialized hardware.
• 5G network slicing – Carriers are also using network virtualization technology to divide up physical 5G infrastructure and create virtual networks with customized features and capabilities for individual customers.

Network virtualization and software-defined networking

Network virtualization is not exactly the same as software-defined networking (SDN), but there are some similarities. They are both valid modern approaches to network management. In fact, some industry experts consider network virtualization, at least its initial iterations, to be a subset of SDN because network virtualization was one of the earliest use cases of SDN. The real difference between them, however, is how the technologies separate resources and functions.

The biggest difference between SDN and network virtualization as it exists today is that SDN doesn’t completely abstract networking functions from physical hardware like network virtualization does. SDN virtualizes mainly the network control plane while still relying on physical routers and switches to move traffic around the network. The physical routers and switches can be operated through software, but the routers and switches do the actual packet routing. SDN still results in networks that are more flexible and agile than traditional networks.

By virtualizing the control plane and abstracting it from the hardware’s data plane, SDN enables networks to be programmable and flexible. The rationale behind separating control from data planes can be explained several ways:

• It gives customers the freedom to use the software and hardware of their choice, and because hardware doesn’t need all the control functions, it can sometimes cost less.
• Users can create extensive networks composed of multiple hardware devices for a wider range of packet forwarding strategies and more precise flow of data traffic.
• Network programmability is a key element of SDN, and SDN allows users to use third-party tools and APIs to get a single point of control or interaction across the entire network.

SDN and network virtualization aren’t competing technologies, but there are instances where one is used more often than the other. SDN is all about centralizing network management where network virtualization is focused more on distributing network functions among multiple virtual hosts. For instance, SDN is typically deployed in data centers or across large, on-premises campuses where centralized control is an advantage. It makes network management easier and more agile. Network virtualization is used more than SDN in WANs, where virtualization can help reduce the need for physical hardware. Because network virtualization can abstracts network functions, it can minimize costs and help reduce WAN complexity.

The two technologies are also sometimes used together in hybrid environments. For instance, you might use SDN to streamline network management and deploy network virtualization (or more specifically NFV) to virtualize specific functions like load balancers or firewalls.

Network virtualization software

To get the most value and benefit out of network virtualization, you need the right tools for management, monitoring, security, troubleshooting, and so on. You have many choices today when it comes to network virtualization tools, so it’s important to do your homework and make sure the solutions you choose will be able to deliver what you need.

There are tools for creating and managing virtual networks, programming and controlling network devices and services, deploying and orchestrating network functions, testing and monitoring virtualized networks, identifying and troubleshooting issues, and much more.

Here are some things to consider when looking for network virtualization software:

• Understand the size, complexity, and topology of your network. A good solution should be able to handle its structure and scale.
• Consider how you use your network now and how you might in the future. What criteria is most important—costs? Support? Security?
• What is the solution provider’s reputation in the industry? Is it respected, with a proven track record?
• What’s the provider’s vision for future improvements or updates of the solution?
• What kind of documentation and support does the provider offer? Is it included in the purchase price?
• How well does the solution integrate with other infrastructure or systems?

With some planning and discussion with your organization’s stakeholders, you can find the network virtualization solution that will meet your needs today and tomorrow.

Related articles:

• What is virtualization?
• What is a virtual machine?
• What is virtual desktop infrastructure (VDI)?
• What is server virtualization?
• What is application virtualization?