Dedicated server rental

If the job fits on a VPS, take the VPS: it costs less, is handed over right after payment and its resources are added in a couple of clicks. A dedicated server is for the point where you run into something a virtual machine does not have.

Usually that is one of three things: a database that needs physical drives under its own load; a project that needs tens of terabytes, which VPS plans simply do not carry; or access to the hardware itself: your own virtualization, your own system image, access to the server outside the operating system.

One more argument is predictability: on a dedicated server every core and all the memory are yours, with no neighbors on the server at all. To compare: Linux VPS plans.

The server itself really is cheaper in the office, but it does not run on its own. It needs bandwidth with headroom, power that survives an outage in the neighborhood, cooling that does not stop over the weekend, a guarded room and somebody who will come at night when something goes down.

A data center already has all of that: redundant power and cooling, round-the-clock security, our own network and our engineers next to the server. A failed drive is replaced by us, not by you.

The money works differently too: rent is a monthly cost instead of buying hardware years ahead, and it does not age at your expense either. Your own server in the office makes sense when regulations keep the data inside the company; in every other case count not the price of the hardware, but the price of everything around it.

Most often, whatever is cramped or uncomfortable on shared resources:

• databases under load that need predictable access to the drives;
• your own virtualization: your hypervisor, your virtual machines, your split of the resources;
• storage of tens of terabytes: the configuration with eight bays takes up to 132 TB;
• large shops and applications that have outgrown a virtual machine;
• file and video processing, builds and tests: anything that likes many cores at once.

Two neighboring services are not on this page: servers with graphics cards and hosting your own hardware, that is colocation. We do have them, but they are sold separately, so ask support about those.

Start not with the price, but with what is actually going to load the server.

• CPU. Each card carries two scores: single-core and multi-threaded. The first matters where the work runs in one thread, the second where there are many threads: virtualization, builds, data processing. A high-frequency server and a many-core server win at different things.
• Memory. Look at the ceiling rather than the base amount: on most configurations it is three to five times higher, and that is how far the server can grow without a move.
• Drives. Capacity is not the only thing that counts; free bays do too: while there are bays, drives are added to a running server. NVMe is faster, SATA Enterprise is cheaper per terabyte.
• Network. The base port is 100 Mbit/s with unmetered traffic and goes up to a gigabit at order time.

The array in the base configuration is software. If a hardware controller is needed, say so before ordering and we will pick a server for it. And if none of the ready configurations fits, we will build another one: that is routine here.

Yes. The term is chosen at order time: a month, a quarter, six months or a year. Setup is free on any term, and the term does not affect the configuration.

The only difference is the price: paying for a year makes the month 10% cheaper than paying monthly. Prices on the pages are shown per month with annual billing; for other terms the client area recalculates the amount at checkout.