VMware ESXi and vSphere Cluster Management

Physical and Virtualized Host Memory in VMware ESXi

Learn how physical, virtual, guest-physical, and host-physical memory work in VMware ESXi, including address translation, VM startup, paging, and memory isolation.

Memory virtualization is the mechanism that lets multiple guest operating systems use one physical server while each guest believes it has its own machine memory. To understand VMware ESXi memory, separate three views: an application's virtual address space, the guest operating system's physical memory, and the host's actual physical RAM.

Memory in a nonvirtualized system

In a conventional, nonvirtualized server, the operating system manages the physical RAM installed in that computer. It decides which parts of RAM are available, which process can use them, and when memory can be reclaimed.

Applications normally request memory through operating-system interfaces. For example, a database process may ask the operating system for additional memory, use that memory, and later release it. The operating system does not normally give the application unrestricted access to RAM addresses.

Instead, each process receives a virtual address space: a range of addresses visible to that process. These addresses are not direct descriptions of locations in the server's RAM. The operating system and processor memory-management hardware translate them into physical memory locations.

This separation is important. If every application could directly address physical RAM, one faulty or hostile process could overwrite another process or the operating system itself. Virtual addressing provides both convenience and protection.

Virtual memory fundamentals

Virtual memory is a memory abstraction supported jointly by the operating system and the CPU's memory-management hardware. It allows software to use virtual addresses while the system translates those addresses to physical memory.

Pages and address translation

Memory is commonly managed in fixed-size units called pages. A virtual page belongs to a process or operating-system virtual address space. A physical page is a fixed-size unit of actual RAM. A memory mapping associates a virtual page with a physical page.

When an application accesses a virtual address, the processor consults translation data prepared by the operating system. The translation identifies the physical page and the offset within that page. The processor's memory-management hardware then performs the conversion as part of the memory access.

Application virtual address
        -> operating-system translation data
        -> processor memory-management hardware
        -> physical page in RAM

The operating system creates and protects the mappings. The processor enforces them while instructions execute. Frequently used translations may be cached by the processor to avoid repeating the full lookup for every access.

Benefits of virtual memory

Capability	How Virtual Memory Supports It	Administrative Significance
Large address spaces	Each process can receive an address range that is independent of the layout of physical RAM.	Applications can be developed without managing raw RAM locations directly.
Process protection	Mappings and permissions restrict one process from reading or writing another process's pages.	A process failure is less likely to corrupt unrelated workloads or the operating system.
File mapping	File contents can be represented within a process's virtual address space.	Applications can work with files through memory-style access where supported.
Paging and swapping	Pages can be managed independently, including movement between available memory locations and secondary storage.	The operating system can handle changing memory demand, although storage-backed memory is much slower than RAM.
Workload isolation	Separate address spaces keep processes' memory views distinct.	Multiple applications can safely share one operating system.

Virtual memory is not unique to virtual machines. Modern operating systems and CPUs commonly use it on physical servers, desktops, and laptops. A VM adds another managed memory layer on top of this existing capability.

Memory layers in an ESXi environment

VMware ESXi is a hypervisor, meaning a virtualization layer that runs virtual machines, manages access to host resources, and enforces isolation. It sits between guest operating systems and the physical server hardware.

A guest operating system is the operating system installed inside a virtual machine. The guest manages processes and application memory as though it were running on its own computer. ESXi, however, controls how the VM's memory view is backed by the server's actual RAM.

Address Space	Visible To	Purpose	Mapped To
Application virtual address space	Processes inside the guest operating system	Provides each application with its own protected range of addresses.	Guest physical memory through guest OS and CPU translation.
Guest physical memory	The guest operating system	Represents the machine memory that the guest believes it controls.	Host physical memory through ESXi's virtualization and mapping mechanisms.
Host physical memory	ESXi and the physical server platform	Represents the actual RAM installed in the server.	Physical RAM hardware.

Virtualization therefore introduces an additional translation requirement. A memory reference may first be translated by the guest's memory-management system and then translated or managed by ESXi before reaching host physical RAM.

How a virtual machine receives memory

When a VM is created or edited, an administrator assigns a memory capacity in the VM's hardware configuration. This value describes the amount of guest physical memory presented to the VM; it is not the size of every application address space inside the guest.

When the VM powers on, ESXi establishes an addressable memory environment of the configured size. The guest sees that memory as contiguous machine memory and can initialize it, divide it among processes, and manage it using its normal operating-system memory mechanisms.

For example, if a VM is configured with 8 GB of memory, the guest operating system is presented with approximately 8 GB of guest physical memory for its own use, subject to the platform's configuration and normal system requirements. A database process inside that guest may have a virtual address space that is smaller, larger in address range, or otherwise different from the VM's 8 GB configured memory.

The VM does not directly own a fixed, independently controlled region of host RAM. ESXi maintains the relationship between guest memory and host physical memory and can manage that relationship according to host conditions and configured resource policies.

Virtual machine memory configuration

Where it is set: During VM creation or in the VM's virtual hardware configuration when the VM is edited.
What it means: The capacity of memory presented to the guest operating system.
What it does not mean: The total virtual memory used by applications or a promise that the guest directly controls particular host RAM addresses.
What the guest sees: An addressable and usable memory range that is designed to appear contiguous to the guest operating system.

Conceptual address-translation flow

Physical server

On a nonvirtualized server, the conceptual path is direct from the application through the operating system's address translation to the server's physical RAM:

Application virtual address
        -> guest or host operating-system translation
        -> processor memory-management hardware
        -> server physical RAM

Virtual machine on ESXi

Inside a VM, the application first uses an address from its process virtual address space. The guest operating system translates that reference into the guest's machine-memory model. ESXi then manages the mapping from that guest physical reference to host physical RAM.

Application virtual address
        -> guest OS and guest memory-management hardware
        -> guest physical memory reference
        -> ESXi virtualization and memory mapping
        -> host physical RAM

The exact implementation can use hardware-assisted virtualization and different mapping techniques, but the conceptual distinction remains: guest physical memory is a view presented to the guest, while host physical memory is the actual RAM controlled by ESXi.

Memory isolation and protection

ESXi can run multiple VMs on one physical host while keeping their memory domains separate. Each VM has its own guest physical memory view, and ESXi controls which host physical pages can back that view.

A VM must not be able to read or alter another VM's memory. The hypervisor enforces this through protected memory mappings and processor-supported access controls. A guest operating system can manage memory within its own VM, but it cannot use ordinary guest memory references to bypass ESXi and access another VM's memory.

This protection works together with isolation inside each guest. The guest OS separates its own processes, while ESXi separates one guest memory domain from another. Both levels are necessary when workloads are consolidated on the same server.

Multiple VMs sharing a host

VM 1: guest memory domain ----\
VM 2: guest memory domain ----- > ESXi mappings -> host physical RAM
VM 3: guest memory domain ----/

The VMs share the host's hardware platform, but sharing hardware does not imply sharing unrestricted memory access. ESXi maintains the boundary between their memory mappings.

Physical server memory versus ESXi VM memory

Aspect	Nonvirtualized Server	Virtual Machine on ESXi
Operating system view of memory	The operating system manages the server's machine memory directly within the platform's controls.	The guest operating system manages guest physical memory presented by ESXi.
Application address space	Applications use virtual addresses translated by the host operating system and CPU.	Applications use virtual addresses translated within the guest memory model.
Translation layers	Application virtual address to physical memory.	Application virtual address to guest physical memory, then to host physical memory.
Physical RAM ownership or control	The physical server operating system controls use of installed RAM.	ESXi controls host physical RAM and maps portions of it to VMs.
Isolation boundaries	Process and operating-system protections separate applications.	Guest OS protections separate processes, and ESXi protections separate VMs.
Ability to run multiple operating systems	Normally one active operating-system instance controls the server.	Multiple independently operating guest systems can share one host.

Common misconceptions and troubleshooting

A VM directly owns a fixed region of host RAM

Misconception: A VM must have an exclusive, permanently assigned range of host physical addresses.

Correction: The guest sees configured memory as its machine memory, but ESXi controls the mapping to actual host RAM. Do not confuse the guest's memory view with direct ownership of host physical addresses.

The vSphere memory value is an application's virtual memory size

Misconception: A VM configured with a particular memory value gives every application an address space of that same size.

Correction: The configured value describes guest physical memory presented to the guest OS. Applications receive their own virtual address spaces through guest operating-system interfaces.

VMs can inspect one another because they share a host

Misconception: Physical co-location means a VM can read another VM's memory.

Correction: ESXi enforces separate memory mappings. A guest's normal memory references remain within its permitted memory domain.

Virtual memory is a feature unique to virtual machines

Misconception: Virtual memory begins when a workload is placed in a VM.

Correction: Virtual memory is already a standard operating-system and CPU capability. ESXi adds virtualization of the guest's machine-memory view and another layer of mapping.

Relationship to other ESXi resources

Memory virtualization is one of three foundational hypervisor capabilities:

CPU virtualization presents virtual processors and schedules their execution on physical CPU resources.
Memory virtualization presents guest physical memory and maps it to host physical RAM while enforcing isolation.
Virtual networking presents virtual network adapters and connects guest traffic to physical or virtual network paths.

Understanding memory layers is foundational for later work in VM sizing, capacity planning, performance monitoring, memory overcommitment, and ESXi memory reclamation. It also helps distinguish a guest operating system's internal memory behavior from the host's physical resource state.

Exam-relevant summary

Physical memory is the actual RAM installed in the server.
Virtual memory gives a process or operating system a virtual address space and uses address translation to reach physical pages.
A virtual page maps to a physical page, subject to operating-system and processor protections.
In an ESXi VM, application virtual memory, guest physical memory, and host physical memory are distinct concepts.
The guest OS manages memory as though it owns a machine, while ESXi controls the mapping to host RAM.
ESXi adds a memory-translation layer between guest physical memory and host physical memory.
Memory isolation prevents one VM from reading or modifying another VM's memory.
The VM memory setting specifies memory presented to the guest OS, not an application's virtual address space and not necessarily a fixed host-RAM region.

For related foundations, review the VMware ESXi online course, installing a guest operating system, and adding virtual hardware to a VM. Memory availability also connects to vSphere admission control when evaluating whether a cluster can safely start and run workloads.