Virtualization Concepts and Hypervisors

Virtualization is the process of creating a software-based (virtual) representation of a physical resource, such as a server, storage device, or network. The primary goal is to abstract the physical hardware so that multiple operating systems and applications can run on a single physical machine, sharing its resources efficiently. Before virtualization, each physical server typically ran a single operating system and application, leading to low utilization (often 5-15% of CPU capacity). Virtualization allows consolidation, reducing hardware costs, power consumption, and physical space. It also enables rapid provisioning, disaster recovery, and isolation between workloads.

At the heart of virtualization is the hypervisor (also called a Virtual Machine Monitor, VMM). The hypervisor is a software layer that sits between the physical hardware and the virtual machines (VMs). It manages access to the underlying hardware resources: CPU, memory, storage, and network. Each VM contains a guest operating system and its applications, and the hypervisor provides each VM with a set of virtual hardware components (virtual CPU, virtual memory, virtual disk, virtual network interface). The guest OS believes it is running on real hardware, but all its instructions are actually processed by the hypervisor, which translates them to the physical hardware.

There are two main types of hypervisors: Type 1 (bare-metal) and Type 2 (hosted).

Type 1 Hypervisor (Bare-Metal): Installs directly on the physical hardware, with no underlying operating system. It acts as a lightweight operating system itself, managing VMs directly. Examples: VMware vSphere/ESXi, Microsoft Hyper-V (when installed as a role on Windows Server, but can also be bare-metal), Citrix XenServer, KVM (Linux kernel-based). Type 1 hypervisors are used in data centers and enterprise environments because they offer high performance, low overhead, and direct access to hardware.

Type 2 Hypervisor (Hosted): Runs on top of a host operating system (e.g., Windows, macOS, Linux). The host OS manages the physical hardware, and the hypervisor runs as an application. Examples: Oracle VirtualBox, VMware Workstation, VMware Fusion, Parallels Desktop. Type 2 hypervisors are common for desktop virtualization, development/testing, and lab environments. They have more overhead because hardware access goes through the host OS, but they are easier to set up and use.

CPU Virtualization: The hypervisor schedules virtual CPUs (vCPUs) on physical CPU cores. Each VM is assigned a certain number of vCPUs. The hypervisor uses time-sharing to give each vCPU access to physical cores. Overcommitment occurs when the total number of vCPUs across all VMs exceeds the number of physical cores. For example, a host with 16 cores might run VMs with a total of 32 vCPUs. The hypervisor manages this by scheduling, but heavy overcommitment can lead to performance degradation. The exact ratio depends on workload; typical production ratios are 4:1 to 8:1 for general workloads.

Memory Virtualization: The hypervisor maps virtual machine memory to physical memory. Each VM has a configured amount of memory (e.g., 4 GB). The hypervisor uses memory ballooning (a driver inside the guest OS that reclaims unused memory) and transparent page sharing (deduplication of identical memory pages) to optimize usage. Overcommitment of memory is more risky than CPU; if the host runs out of physical memory, it may use swap (disk), causing severe performance issues. Production environments often avoid memory overcommitment or use it conservatively.

Storage Virtualization: Virtual disks are stored as files on a storage system (local disk, SAN, NAS). The hypervisor provides a virtual storage controller (e.g., SCSI, SATA, NVMe) to the VM. Thin provisioning allows the virtual disk file to start small and grow as data is written, up to a maximum size. Thick provisioning allocates all space upfront. Snapshots create a delta file that tracks changes since the snapshot point; they can cause performance loss if left for long periods.

Network Virtualization: Each VM has one or more virtual NICs (vNICs) connected to a virtual switch. The vSwitch can be configured with VLANs, bandwidth limits, and security policies. The hypervisor manages MAC addresses for vNICs. For external connectivity, the vSwitch is connected to a physical NIC via a uplink. Advanced features include SR-IOV (Single Root I/O Virtualization), which allows a VM to directly access a physical NIC for near-native performance.

Virtualization runs multiple OS instances on one host, each with its own kernel. Containers share the host OS kernel but isolate applications at the process level. Containers are lighter and start faster, but are less isolated than VMs. The CompTIA A+ exam focuses on VMs, but you should know that containers are a different technology (e.g., Docker).

Virtualization is the foundation of cloud computing. Cloud providers (IaaS) use Type 1 hypervisors to offer virtual machines to customers. The exam objective 4.1 covers cloud concepts like IaaS, PaaS, SaaS, and deployment models (public, private, hybrid). Virtualization enables multi-tenancy, elasticity, and self-service provisioning in the cloud.

For Type 2 hypervisors like VirtualBox, you can use the command line:

VBoxManage list vms
VBoxManage startvm "VM Name" --type headless
VBoxManage controlvm "VM Name" poweroff

For VMware Workstation:

vmrun start "C:\path\to\vm.vmx"
vmrun list
vmrun stop "C:\path\to\vm.vmx" soft

For Hyper-V (PowerShell):

Get-VM
Start-VM -Name "VMName"
Stop-VM -Name "VMName"

For ESXi (CLI via SSH):

esxcli vm process list
vim-cmd vmsvc/getallvms
vim-cmd vmsvc/power.on <vmid>

In a typical enterprise data center, virtualization is used to consolidate workloads. For example, a company might have 20 physical servers running different applications (web server, database, file server, etc.) each at 10% CPU utilization. By virtualizing, they can run those 20 workloads on 2-3 physical hosts running VMware ESXi, achieving 70-80% utilization. The hypervisor provides HA (High Availability) so that if a host fails, VMs are automatically restarted on another host. Storage is often shared via SAN (Fibre Channel or iSCSI) or NAS (NFS) to enable live migration. For instance, a DBA might need to move a database VM to a host with more memory for a maintenance window; using vMotion, the VM moves with no downtime.

Another scenario: A development team uses Type 2 hypervisors like VirtualBox on their laptops to create isolated environments for testing software. Each developer runs multiple VMs with different OS versions (Windows, Linux) to test compatibility. They use snapshots to quickly revert to a clean state after testing. However, they must be careful not to overcommit memory on their laptops, as running 4 VMs with 2 GB each on a laptop with 8 GB RAM will cause swapping and sluggish performance.

A common problem: An administrator creates a VM with thin-provisioned virtual disks and takes multiple snapshots without monitoring disk space. The datastore runs out of space, causing all VMs on that datastore to pause or crash. The solution is to set alarms on datastore usage and regularly consolidate snapshots. Another issue: Overcommitting CPU without monitoring can lead to CPU ready time (time a VM is ready to run but waiting for a physical CPU), degrading performance. The admin must use performance monitoring tools (e.g., vCenter performance charts) to detect and adjust allocations.

In cloud environments like AWS or Azure, the underlying hypervisor is abstracted from the customer. The customer provisions VMs (EC2 instances) without knowing the hypervisor type, but the cloud provider uses Type 1 hypervisors (e.g., Xen, KVM) to manage multi-tenancy. The customer can choose instance types with different vCPU and memory ratios, similar to configuring a VM on-premises.