VPS vs VDS vs dedicated server: how to choose

A virtual private server is a virtual machine on a physical host that runs several virtual machines side by side. A hypervisor divides the host's processors, memory and storage between them and keeps each one isolated: you get your own operating system, root or administrator access, your own IP address and the resources your plan allocates.

Because the host is shared, providers can allocate more virtual CPUs in total than the host has physical threads, on the assumption that not every server is busy at the same moment. This is normal practice and keeps prices low. It suits workloads that are quiet most of the time and busy in short bursts, such as most websites, development environments and small APIs.

What sharing means in practice

When several virtual machines want the CPU at the same time, each waits a little. On Linux you can see this as steal time: the st value in top shows the share of time the hypervisor gave your virtual CPU to another machine. Occasional low values are harmless. Steal time that stays high during your busy hours means your workload needs reserved resources.

VDS stands for virtual dedicated server. Across the industry the word is used loosely: some providers mean larger allocations, others CPU capacity that is not shared with other customers. Always read the plan details rather than the label: how many virtual CPUs you get, how much memory, what kind of storage, and which features the server exposes.

At HyperDC the rule is precise: we call a server a VDS only when nested virtualization is available on it, so you can run your own virtual machines, emulators or a hypervisor inside it. Our Linux, Windows and macOS VDS plans offer it, together with larger CPU, RAM and NVMe storage allocations than our VPS plans. They suit production websites and shops, databases, build servers, game servers and virtualization labs.

Why Apple Silicon macOS is a VPS

On Apple silicon Macs, macOS virtual machines run on Apple's Virtualization framework, which does not offer nested virtualization for macOS guests. Because you cannot run virtual machines inside them, our Apple Silicon macOS servers are sold as Apple Silicon macOS VPS. They still give you a complete macOS system with administrator access for Xcode builds and app testing.

A dedicated server gives you the whole physical machine. Nothing else runs on it, so performance is as consistent as the hardware allows, and you can use what a virtual machine cannot offer: your own disk layout, a remote management interface where the plan includes one, or a hypervisor of your own.

Typical reasons to move to bare metal

• Sustained load: databases, analytics or game servers that keep the CPU or disks busy for hours.
• Licensing: some software is licensed per physical core or per processor. Windows Server, for example, is licensed per physical core with minimums per processor and per server, which can favour one well-used dedicated machine.
• Isolation: virtual machines are isolated by the hypervisor, but a single-tenant server makes it simpler to show auditors that no one else shares your hardware.
• Hardware access: running your own virtualization platform, using specific disks or controllers, or needing full control of the network stack.

The trade-off is flexibility. Adding memory or disks to a physical server is a hardware job, and moving to a bigger machine usually means a migration, so plan capacity with some headroom.

Some projects need to run their own virtual machines, for example to test hypervisors, run Android emulators or build a small lab. That requires nested virtualization: the virtual server passes the processor's virtualization features on to your system. At HyperDC it is available on every VDS plan and on nested dedicated servers, which add larger dedicated resources for heavier labs, without renting a whole physical machine. VPS plans, including Apple Silicon macOS VPS, do not offer it. Expect some performance overhead compared with running directly on hardware.