5.3.2: Virtualization of Cloud Infrastructure

Virtualization is the technological principle that enables cloud computing by abstracting computing resources from the underlying hardware. In the context of cloud strategy, virtualization of cloud infrastructure means that the enterprise should design its cloud environment to maximize resource flexibility and utilization through virtual layers. This principle covers both hardware virtualization and software virtualization techniques that the cloud uses to pool resources.

• Hardware virtualization refers to using a hypervisor or similar technology to turn physical hardware (servers, storage, networks) into configurable virtual machines or virtual resources. In a cloud, hardware virtualization allows dynamic creation of server instances, on-the-fly allocation of storage, and plug-and-play addition of hardware resources. For example, an enterprise can provision a new virtual server in minutes on a cloud provider without worrying about the physical server installation – the cloud’s infrastructure management handles it in a virtualized manner.
• Software virtualization can refer to techniques like containerization or using software images to encapsulate applications. It involves using software-based abstraction (such as virtual appliances, containers like Docker, or application virtualization) to enable sharing and portability of software on the cloud. Software image management means you can create a template of an application environment and deploy it multiple times consistently on virtual infrastructure.
  

By following the virtualization principle, an enterprise ensures that its cloud strategy leverages the key cloud benefit of resource abstraction. Cloud resources (compute, storage, networking) are treated as elastic pools rather than fixed hardware. This has several strategic implications: scalability (applications can scale out by spawning new virtual instances as load grows), efficient utilization (multiple virtual machines share the same physical hardware safely, increasing overall utilization), and isolation (workloads separated in VMs or containers for security and stability).

From a design standpoint, virtualization is an extension of the enterprise’s existing architectural layers. In fact, researchers note that the virtualization principle in cloud is an extension of the operational system layer in a Service-Oriented Architecture (SOA) stack. It inherits the idea of separating the logical computing environment from physical hardware, which aligns with enterprise architecture practices (like in traditional data centers using VMware or Hyper-V for virtualization). A modern cloud strategy also includes containerization (e.g. Kubernetes) as part of virtualization – treating containers as the new unit of compute that can be flexibly scheduled on underlying machines.

Example: Netflix is often cited as a real-world example of virtualization and cloud at scale. Netflix migrated from physical data centers to AWS cloud infrastructure, leveraging virtualization to run thousands of Linux server instances on AWS’s hardware. By doing so, Netflix achieved elastic scaling – when user streaming demand surges, Netflix can launch dozens of new virtual servers across AWS regions, then tear them down when no longer needed. This elasticity is possible only because the cloud abstracts servers as software-defined resources. Moreover, Netflix uses container technologies and virtual networking extensively to deploy its microservices globally. The virtualization principle in Netflix’s strategy has allowed it to focus on innovation at the software layer while leaving capacity management to the cloud provider.

In summary, virtualization of infrastructure is a foundational principle for any enterprise cloud strategy, as it underpins the cloud’s ability to provide computing resources on-demand. When designing your strategy, ensure that your architectures (whether IaaS virtual machines, PaaS containers, or even serverless functions) take full advantage of virtualization — this typically means designing applications to be decoupled from any single physical hardware and capable of running in a distributed, virtual environment.

---