Explore our new Cybersecurity Terms Glossary, explained by our experts.
Cybersecurity Terms Glossary
Hypervisor security concerns the layer that enforces isolation between virtual machines: its own attack surface, the escape vulnerabilities that break that isolation, and the management plane that controls every guest.

What Is Hypervisor Security?
Every argument for consolidating workloads onto shared hardware rests on one assumption: that the hypervisor keeps guests apart. Hypervisor security is the discipline of testing that assumption. It covers the attack surface the hypervisor exposes to the virtual machines it hosts, the vulnerabilities that let a guest break out into the host, and the management plane that can create, copy, or destroy every workload in the estate.
A type 1 hypervisor runs directly on the hardware. VMware ESXi, Microsoft Hyper-V, KVM, and Xen are the common examples, and they carry a comparatively small trusted computing base. A type 2 hypervisor runs as an application on a general-purpose operating system, which inherits every weakness of that host. In production, type 1 dominates and the interesting attack surface is not the scheduler or the memory manager but the device model: the emulated network cards, storage controllers, USB controllers, graphics adapters, and clipboard and file-sharing services that the hypervisor offers each guest.
That is where escapes come from. VENOM, CVE-2015-3456, was a buffer overflow in the QEMU virtual floppy disk controller, reachable even when no floppy device was attached, and it affected Xen and KVM deployments across the hosting industry. Pwn2Own has produced VMware Workstation and Hyper-V escapes repeatedly through graphics and device emulation paths. The pattern is consistent: the more devices you expose to a guest, the more code the hypervisor must parse on that guest’s behalf.
Attackers rarely bother with an escape when the management interface is easier. The vCenter and ESXi vulnerabilities of recent years have been exploited at scale for exactly that reason.
Encrypting the datastore of a single host destroys dozens of virtual machines at once and defeats most in-guest backup agents. That economics is why ransomware groups build dedicated ESXi encryptors and why hypervisor hardening has moved from an infrastructure concern to a board-level one.
The measures that matter are mostly about reachability and separation rather than exotic configuration.
Put management interfaces on a dedicated network that is not routable from user subnets and never from the internet, and require a jump host with multi-factor authentication to reach it. Enable lockdown mode on ESXi so that direct host access is restricted, and disable services such as SSH and SLP unless they are needed for a defined purpose. Break the dependency between the hypervisor and the domain: an Active Directory compromise should not hand over every virtual machine, so use local or separately governed accounts for host administration.
Reduce the device model. Remove virtual floppy, serial, parallel, and USB controllers from templates, disable clipboard and drag-and-drop between guest and host, and switch off unused paravirtual services. Keep Secure Boot enabled for the hypervisor itself, use a vTPM for guests that need measured boot, and apply patching to hypervisor hosts on the same urgency as internet-facing systems, because that is effectively what the management plane has often become.
Finally, treat the host as a monitored asset. Hypervisor logs, authentication events, and datastore activity should reach your SIEM, and backups of virtual machines must be immutable and stored outside the virtual estate they protect.
In a public cloud, the hypervisor belongs to the provider and the shared responsibility model places escape risk on their side of the line, which is why side-channel and speculative execution research attracts such attention from cloud security teams. On premises, the entire layer is yours, including the firmware beneath it. For broader programme concerns across the virtual estate, see our note on virtualisation security.
Our Infrastructure and Architecture Review Services assess hypervisor configuration, management plane exposure, and backup isolation together, since an attacker who reaches one usually reaches all three.
It is an attack in which code running inside a guest virtual machine breaks the isolation boundary and executes on the hypervisor or host. Most escapes exploit the emulated devices a hypervisor presents to guests, such as storage, network, or graphics controllers, rather than the core memory management. They are rare and valuable, which is why reducing the virtual hardware exposed to a guest is effective.
Because the return is enormous for one intrusion. Encrypting the datastore on a single host destroys every virtual machine running on it, often dozens, and in-guest backup agents and endpoint protection cannot act once the underlying storage is encrypted. Several groups maintain purpose-built Linux encryptors for ESXi, and most campaigns begin with an exposed or unpatched management interface.
It is convenient and it concentrates risk. Domain integration means that compromising the directory also compromises the virtual estate, which CVE-2024-37085 demonstrated through a group name that could simply be created. If you do integrate, restrict which groups receive administrative rights, monitor for group creation, and keep a break-glass local account that is not dependent on the domain.
No. In infrastructure as a service, the provider operates the hypervisor and is responsible for isolation between tenants, while you remain responsible for the guest operating system, its configuration, and the identity controls around the management API. The practical equivalent of hypervisor hardening in that model is tightening the cloud control plane and the permissions attached to it.
You May Be Interested In These
We use a few cookies that are needed to make this website work, and they are always on. We would also like to set analytics cookies, to see which pages are useful; functional cookies, so that our live chat works; and advertising cookies, to measure our campaigns. We will not set any of those until you say yes, and you can change your mind at any time from the Cookie Settings link in the footer.
Choose the cookies you are happy for us to use. Strictly necessary cookies are always on; everything else is off until you turn it on.
These cookies keep the website working: they hold your security checks, your form submissions and the choice you make here. The site cannot run without them, so they are always on.
These let our live chat open and remember your conversation, and let the newsletter sign-up work. Switch them off and the chat window will not appear.
These tell us which pages are read and how people find us, so that we can improve the site. The reports are about groups of visitors, never about you personally.
These let us measure our advertising and show you our adverts on other websites, such as LinkedIn, Google and Meta. Switch them off and you will still see adverts, but they will be less relevant.