Snapshots
A restore point taken before an upgrade. Rolling back takes minutes. A snapshot usually sits on the same storage, so it is not a backup; see the 3-2-1 backup rule.
Virtualization is a software layer, the hypervisor, that splits one physical server into several isolated virtual machines, each with its own operating system, kernel, RAM, disk and network card. Hosting providers use Type 1 hypervisors such as KVM, so every VPS is one guest on a larger host with a guaranteed share of its resources.
For site owners, developers and students who want to understand what sits under a VPS or cloud server before choosing one.
Last editorial update: 14 November 2026. Prices follow the dated sources on this page; checkout confirms current charges.
Virtualization is a software layer that lets one physical computer run several isolated virtual machines (VMs), each with its own operating system, kernel, memory, disk and network card. The physical machine is the host; the VMs are guests. Each guest behaves as if it were a separate server.
The layer that does the splitting is the hypervisor. It hands every guest a defined share of CPU time, RAM and storage, and stops one guest reading another's memory. That is the technology behind every VPS and every public cloud: a VPS is simply one guest on a larger host.
| Term | Meaning | Hosting example |
|---|---|---|
| Host | The physical server with real CPUs, RAM and disks | A node in the provider's data centre |
| Guest / VM | A virtual computer running on the host | Your VPS |
| Hypervisor | Software that creates, schedules and isolates VMs | KVM, VMware ESXi, Hyper-V, Xen |
| Image | A template disk used to boot a new VM | Ubuntu 24.04 or Windows Server image |
| Snapshot | A saved point-in-time state of a VM | "Before PHP upgrade" restore point |
A Type 1 (bare-metal) hypervisor runs directly on the hardware and is what hosting providers use; a Type 2 hypervisor runs as an application on a desktop operating system and suits testing on a laptop.
Modern hypervisors rely on CPU extensions, Intel VT-x and AMD-V, so guests run most instructions at near-native speed. Paravirtualised virtio drivers do the same for disk and network, which is why a well-configured KVM guest performs close to bare metal for typical web workloads.
Full virtualization gives each guest its own kernel; container virtualization lets all guests share the host's kernel. That one difference decides which operating systems you can run, how strong the isolation is and how densely a host can be packed.
| Attribute | KVM / full VM | OpenVZ / LXC container | Docker container |
|---|---|---|---|
| Kernel | Own kernel per guest | Shared host kernel | Shared kernel of the machine it runs on |
| Operating systems | Linux, Windows, BSD | Linux only, host-compatible | Linux userlands (Windows containers need a Windows host) |
| Custom kernel modules, VPN, Docker inside | Yes | Limited, depends on host settings | Not applicable; it is the app layer |
| Boot time and overhead | Seconds to a minute; more RAM overhead | Very fast, very light | Sub-second, lightest |
| Typical use | VPS and cloud servers | Budget container VPS, system containers | Packaging apps inside a VM or server |
Docker is not a replacement for a VPS; it usually runs inside one. See what a Docker container is for the application side.
The hypervisor enforces separate memory address spaces, schedules CPU time per guest, gives each VM its own virtual disk and network interface, and can cap I/O per guest. A crash or runaway process in one VM stays inside that VM.
st column in top).Isolation is strong but not absolute. Shared physical storage and CPU caches can still carry "noisy neighbour" effects, which is why host density matters as much as the hypervisor.
Because a VM is essentially files plus a running process, it can be paused, captured as a snapshot, cloned or moved to another host while it keeps running.
A restore point taken before an upgrade. Rolling back takes minutes. A snapshot usually sits on the same storage, so it is not a backup; see the 3-2-1 backup rule.
The hypervisor copies a running VM's memory to another host and switches over with a pause of well under a second, so hardware maintenance need not reboot your server.
vCPUs, RAM and disk can be increased by changing the VM's allocation, usually with one reboot, instead of buying new hardware.
A configured VM can be saved as an image and used to launch identical servers for staging or scaling.
Virtualization is the technology that splits hardware into VMs; cloud computing is the service built on top of it, adding self-service provisioning, APIs, pooled capacity and failover across many hosts. Every cloud server is a virtual machine, but a single VPS on one host is not automatically a cloud.
In practice the difference shows up in what happens when a host fails or you need capacity quickly. A cloud platform can restart your VM on another node from shared storage; a standalone VPS waits for its host to be repaired. The trade-offs are compared in cloud vs VPS, and WebJi® cloud hosting starts at the Cloud Entry tier of 2 vCPU, 8 GB RAM and 100 GB storage for ₹5,900 a month before 18% GST.
On a Linux VPS, systemd-detect-virt prints the virtualization type in one word, such as kvm, openvz, lxc or none for bare metal. Two more commands confirm the details.
# virtualization technology in use
systemd-detect-virt
# hypervisor vendor and CPU model as the guest sees it
lscpu | grep -iE 'hypervisor|model name'
# on a physical host: does the CPU support hardware virtualization?
grep -cE 'vmx|svm' /proc/cpuinfo # a number above 0 means yes
Before buying, ask the provider four questions: is it KVM or a container, is RAM guaranteed or burstable, how are disk I/O and CPU shared per node, and are snapshots included. A WebJi® Linux VPS is quoted in INR before payment, and what VPS hosting is covers plan selection.
Slightly. With hardware extensions and virtio drivers, a KVM guest typically runs web and database workloads close to bare-metal speed. Noticeable slowdowns usually come from an oversold host, visible as CPU steal time or slow disk, not from the hypervisor itself.
Yes. Because KVM gives each guest its own kernel, the same Linux host can run Windows Server guests alongside Linux ones, provided the Windows licence is covered. Container platforms such as OpenVZ or LXC cannot do this.
Running a hypervisor inside a VM, for example KVM or Hyper-V inside your VPS. It needs the host to expose VT-x or AMD-V to the guest, which many providers disable by default. Docker does not need it, but Android emulators and lab clusters often do.
Not by default: a dedicated or bare-metal server gives you the whole physical machine. You can install your own hypervisor such as Proxmox on it and run your own VMs, which some agencies do to split one server between clients.
Last editorial update: 14 November 2026. Prices follow the dated sources on this page; checkout confirms current charges.
Join 12,000+ businesses across India and overseas. INR pricing, 18% GST shown before payment and 24/7 email support in Hindi and English — average first reply under 60 minutes.