A VPS, or virtual private server, is a form of multi-tenant cloud hosting in which virtualized server resources are made available to a user over the internet through a cloud or hosting provider.
Each VPS is installed on a physical machine, operated by the cloud or hosting provider, that runs multiple VPSs. But while the VPSs share a hypervisor and underlying hardware, each VPS runs its own operating system (OS) and applications and reserves its own portion of the machine's resources (for memory, compute, etc.).
A VPS offers levels of performance, flexibility and control somewhere between those offered by multi-tenant shared hosting and single-tenant dedicated hosting. While it might seem counterintuitive that the multi-tenant VPS arrangement would be called “private”—especially when single-tenant options are available—the term “VPS” is most commonly used by traditional hosting providers to distinguish it from shared hosting, a hosting model where all of the hardware and software resources of a physical machine are shared equally across multiple users.
At the other end of the continuum, some cloud providers (including IBM) offer a level of hosting isolation (and privacy) beyond a multi-tenant cloud server. Two common models include dedicated hosts and dedicated instances. In both of these models, the user is getting access to virtual resources, and is likely taking advantage of a managed hypervisor, but is doing so on dedicated, single-tenant hardware.
The next sections provide more detailed comparisons of VPS, shared and dedicated hosting.
When considering use cases for virtual servers, differences between providers can be truly significant. For traditional hosting providers, a VPS represents a nice balance of cost, flexibility, scalability and control between shared and dedicated hosting. These qualities make it a good fit for eCommerce, apps that have moderate or spiky traffic, email servers, CRM, etc.
But beyond that, virtual servers from major public cloud providers are more robust and feature-rich—they are the foundational building block for much of what is considered “cloud” today and can handle a much more diverse set of workloads.
Connect and integrate your systems to prepare your infrastructure for AI.
Register for the guide on app modernization
Building on the concept of “tenancy,” the most common analogy to the differences between shared, VPS and dedicated hosting are the differences between types of actual housing:
Shared hosting is the most basic, most cost-effective form of hosting. In shared hosting, the resources of one physical machine are made available to all tenants in equal proportions. Shared hosting is ideal for basic, personal websites and web apps that have little traffic, few technical requirements and limited performance or security requirements.
In a shared hosting model, because all tenants are allocated a finite amount of an individual server’s capacity, providers do not allow websites to scale beyond the limits of the plan. Nevertheless, shared hosting is the model most susceptible to the “noisy neighbor” phenomenon—in which certain tenants’ applications unexpectedly consume more than their allotted share of resources, causing performance problems for other tenants. (For more information about shared hosting, see “What is Cloud Hosting?” and "Web Hosting: An Introduction.")
As already noted, VPS hosting is considered a premium option compared to shared hosting. In VPS hosting, shared resources are made available to a user who has greater control over system specifications, guest operating systems and the overall software stack than is the case in shared hosting.
It’s important to note that while VPS hosting exists between shared and dedicated hosting when it comes to control, price and simplicity, it is the most scalable of the three models, and is the closest relative of the VMs/virtual servers offered by most public cloud providers.
Unlike shared and VPS hosting, dedicated hosting offers users access to all the hardware resources of a given server. Dedicated hosting offers the greatest levels of isolation, security, performance and control, in comparison to VPS and shared hosting.
Dedicated hosting is also the most expensive of the three models because of the level of hardware resources allocated to a single customer. It’s also more cumbersome to scale than VPS because scaling requires the provider to configure and provision new, physical hardware resources.
The term “bare metal servers” is sometimes used interchangeably with “dedicated servers,” but providers offering bare metal typically offer more cloud-like characteristics in their dedicated servers—such as provisioning in minutes (versus hours), billing in hourly increments (versus monthly) and providing higher-end hardware, including graphic processing units (GPUs). (See “Dedicated and Bare Metal Servers Explained” for a full exploration into the two options.)
A VPS is commonly understood as a single, virtual machine on a piece of physical hardware that’s shared with other VMs. Dedicated instances and hosts bring another level of isolation, control and visibility to VPS hosting by placing the virtual machines on single-tenant, dedicated hardware.
A dedicated host rents an entire physical machine and maintains sustained access to and control over that machine, its hardware, and whatever software is installed on it. This model provides the maximum amount of hardware flexibility and transparency, workload control and placement, and also offers some advantages for certain bring-your-own license software.
A dedicated instance offers the same single-tenant isolation and control over workload placement, but isn’t coupled with a specific physical machine. That means that it’s an instance. So, for example, if a dedicated instance is re-booted, it might wind up on a new physical machine—a machine dedicated to the individual account, but nonetheless a new machine, potentially in a different physical location.
Dedicated hosts and dedicated instances vary slightly in their management options, pricing models and visibility.
Public, dedicated, reserved and transient “spot” virtual servers with Intel® Xeon® processors.
100% dedicated, single-tenant, bare metal servers.
Deploy secure, highly available clusters in a native Kubernetes experience.
Cloud computing wouldn’t be possible without virtualization. Virtualization wouldn’t be possible without the hypervisor. This thin layer of software supports the entire cloud ecosystem.
Discover what dedicated and bare metal servers are, how they compare to virtual servers and what use cases they best support.
See how cloud hosting makes applications and websites accessible using cloud resources and a network of connected virtual and physical cloud servers.