Hypervisor

What is a hypervisor?

A hypervisor (or virtual machine monitor) creates and runs virtual machines by presenting each with virtualized hardware while multiplexing the real CPU, memory, and devices underneath. It’s the enforcement layer that lets many isolated guests share one physical machine safely — and it’s the thing you never see but always depend on, because every cloud instance you launch is a guest under someone’s hypervisor.

Type 1 vs Type 2

Type 1 (bare-metal) hypervisors run directly on hardware — no host OS beneath them — which is what every serious cloud and datacenter uses: KVM (Linux’s built-in hypervisor, the basis of most cloud infrastructure including AWS’s Nitro-era stack), VMware ESXi, Microsoft Hyper-V, Xen. Type 2 (hosted) hypervisors run as an application on a normal OS — VirtualBox, VMware Workstation — fine for developer laptops, too slow and less isolated for production. The distinction matters because Type 1’s thin, direct-to-hardware design is both faster and a smaller attack surface.

The performance and security angles worth knowing

Modern hypervisors lean on hardware virtualization extensions (Intel VT-x, AMD-V) so guest instructions run near-native rather than being emulated — the reason cloud VMs feel like real machines. AWS’s Nitro system pushed this further, offloading virtualization to dedicated hardware so nearly all the host CPU goes to guests. The security frontier is VM escape — a guest breaking through the hypervisor to the host or sibling VMs — the rare-but-catastrophic vulnerability class (and the reason multi-tenant isolation is taken so seriously). Lightweight hypervisors like Firecracker shrink the hypervisor itself to minimize that attack surface while booting microVMs in milliseconds.

What people get wrong

• Ignoring the hypervisor as “not my layer” — its scheduling and noisy-neighbor behavior directly shape your instance’s real performance.
• Assuming perfect isolation: strong, not infinite; VM escapes exist, which is why sensitive workloads sometimes demand dedicated hosts.
• Confusing hypervisors with containers — different virtualization layers (hardware vs OS), different isolation guarantees.

Primary source: Red Hat — What is a hypervisor?

Full Virtualization vs. Paravirtualization

Hypervisors implement virtualization through different techniques with different performance tradeoffs: full virtualization presents guest operating systems with a complete emulation of standard hardware, requiring no modification to the guest OS but incurring more overhead from that emulation layer, while paravirtualization requires the guest OS to be aware it’s running virtualized and cooperate with the hypervisor directly, reducing overhead at the cost of requiring specially modified guest operating systems. Modern hardware-assisted virtualization (Intel VT-x and AMD-V) has largely closed the performance gap that motivated paravirtualization in the first place, which is why full virtualization is now the dominant approach in most cloud environments, offering broad OS compatibility without the meaningful performance penalty it once carried.