Dedicated Server Hosting

What is a dedicated server and how is it different from a virtual server?

A dedicated server is a real machine reserved for a single customer; the operating system runs directly on the hardware, with no hypervisor shared with other tenants in between. With a VPS or VDS, one physical machine is split into several virtual servers with KVM; on a dedicated server, the machine's CPU cache, memory channels, disk controller and network card serve only your workload.

Who needs a dedicated server?

A dedicated server is needed when even the largest virtual server plan is not enough, or when you need isolation or control at the hardware level. Typical users:

• Large databases: MySQL, PostgreSQL or MSSQL installs that need hundreds of gigabytes of buffer memory and heavy write loads.
• Teams building their own cloud: agencies and IT teams that want to split the server into their own virtual machines with Proxmox VE, VMware ESXi or Hyper-V.
• Hosting companies: providers hosting hundreds of accounts with cPanel/WHM or Plesk.
• Game networks: communities that run several Minecraft, FiveM or Rust servers on one machine.
• Licensing requirements: enterprise software licensed per physical core or socket.
• Organisations with compliance needs: security policies that require hardware not to be shared with other tenants.

If none of these apply, a VDS server with dedicated cores covers most workloads with more flexibility and a lower entry cost. The VDS vs dedicated server page covers the decision in detail.

Which processor and hardware options are available?

LIONFIN dedicated servers come in three hardware classes for different workloads: single-socket Intel Xeon E-series servers, high-clock AMD Ryzen 9 servers and a dual-socket Intel Xeon Gold platform. Core count, memory and number of drives for each plan are listed in the table.

• Xeon E series: a stable, ECC-capable and efficient platform; a balanced choice for web servers, medium-sized databases and business applications.
• Ryzen 9: high single-core clock speed; stands out for game servers, build jobs and applications bound to a single thread.
• Dual Xeon Gold: many cores and large memory capacity; for virtualisation, large databases and parallel jobs.

Memory on every plan is ECC: it corrects single-bit errors in memory cells at the hardware level, preventing silent data corruption and unexplained crashes. On servers that run continuously for long periods, this is the most important stability difference from desktop-class memory.

How should the disks be configured, and do you need RAID?

Dedicated servers have several NVMe drives; combining them with RAID keeps your data and service up if a single drive fails. RAID is not a backup, it is a continuity measure: an accidentally deleted file or ransomware is reflected on all drives at once.

On Linux, software RAID with mdadm is a common and fast method on NVMe drives; ZFS combines RAID, compression and snapshots in a single layer. You can state which layout you want at installation via a support ticket.

How do you access and manage the server remotely?

You connect to a dedicated server through the operating system over SSH (Linux) or Remote Desktop (Windows); models with IPMI access stated on the plan card also have a management interface independent of the operating system. IPMI (Intelligent Platform Management Interface) is a separate management controller on the motherboard that provides:

• Remote power off, power on and reboot when the server hangs.
• A remote console (KVM over IP) where you can see the BIOS and boot screens.
• Operating system installation from an ISO file via virtual media.
• Monitoring of temperature, fan and power supply sensors.

Because the IPMI interface is so powerful, it must never be left open to the public internet; use it with a strong password and, if possible, from specific IPs only. On models without IPMI, if you cannot reach the operating system, open a support ticket from the customer panel and ask the NOC team for a reboot or console check. The first-login and hardening steps for virtual servers apply equally to dedicated servers.

Which operating systems and platforms can you install?

You can install a general-purpose operating system or a bare-metal hypervisor directly on a dedicated server; the choice depends on whether you will use the server for a single job or split it into your own virtual machines.

• Linux distributions: long-term-support releases such as Ubuntu LTS, Debian or AlmaLinux, for web, database and container workloads.
• Windows Server: for Active Directory, MSSQL, IIS and Windows-based business applications.
• Proxmox VE: an open-source virtualisation platform based on KVM and LXC, with a web interface for managing virtual machines and backups.
• Hyper-V: Microsoft's virtualisation, included with Windows Server.

If you want a platform that is not in the operating system list on the order screen, you can install it yourself from virtual media on IPMI models or request installation by support ticket. If you will use Windows Server on a dedicated server, remember that licensing depends on core count.

How are network connectivity and DDoS protection provided?

Dedicated servers connect to the LIONFIN network, with 600 Gbit/s of total upstream capacity, through ports whose speed depends on the plan; port speed and monthly traffic quota are listed in the plan table. Port speed determines how much data your server can send and receive at any moment; the traffic quota limits the total data transferred over the month.

All incoming traffic passes through the RouteFence firewall layer. RouteFence filters volumetric and protocol-based DDoS attacks at the network level before they reach your server. This protection matters especially for game networks and public services; attack traffic is stripped out before it can fill the port.

Network-level filtering does not make a firewall inside the server unnecessary. Keep only the necessary ports open with nftables or UFW on Linux and Windows Defender Firewall on Windows; open management ports (SSH, RDP, IPMI) only to IP addresses you trust whenever possible.

What data center conditions does the server run in?

Dedicated servers run in a data center with a hot/cold aisle rack layout, where the cold aisle is kept between 18 and 24 °C. Since the whole machine is yours, cooling quality is reflected directly in your performance: high-core-count CPUs and many NVMe drives generate significant heat under load, and poor cooling leads to throttled clock speeds and shorter hardware life. See our data center page for more.

The infrastructure is monitored 24/7 by two teams: the NOC tracks network, power and hardware status, and the SOC tracks security events. Our uptime target is 99.99%; this is a target, not a contractual guarantee.

The hardware belongs to LIONFIN; if a disk, memory module or power supply fails, simply open a support ticket from the customer panel. Placing your own hardware in a data center is a different model, known as colocation.

How do you choose the right dedicated server?

Choose a dedicated server by your workload's bottleneck: core count and clock speed for CPU-heavy jobs, memory and disk for databases, port speed and traffic for network-heavy services. Steps:

1. Measure the bottleneck: watch CPU, memory, disk wait (iowait) and network usage on your current server at its busiest hour.
2. Single-core or multi-core? Game servers and some older applications depend on single-core speed; virtualisation and parallel jobs depend on core count.
3. Memory: if a database's hot data fits in memory, disk access drops sharply.
4. Disk layout: calculate the usable capacity left after RAID 1 or RAID 10.
5. Network: compare your monthly traffic and peak bandwidth with the port speed.
6. Licensing: with software licensed per core, unnecessary cores can raise costs.

Concrete hardware values are in the plan table on this page. Scaling a dedicated server is not instant as it is on a virtual server; it usually requires moving to a new server. So be generous with growth headroom in your first choice.

How do you monitor and tune dedicated server performance?

Because you access the hardware directly, you can monitor and tune details you cannot see on a virtual server, such as disk health, CPU temperature and memory placement. Used well, this gets noticeably more out of the same hardware.

• Disk health: track NVMe wear level, temperature and error counters with smartctl -a /dev/nvme0 or nvme smart-log; if you use software RAID, feed the output of cat /proc/mdstat into your monitoring.
• CPU frequency: on Linux the power profile (cpupower frequency-info) can be set to performance mode for latency-sensitive workloads.
• NUMA: on dual-socket servers each processor has its own memory region; pinning a large database or virtual machines to a single NUMA node shortens memory access times. numactl --hardware shows the layout.
• Sensors: on IPMI models you can read fan and temperature values with ipmitool sensor.

Feeding these values into a monitoring stack such as Prometheus and Grafana and defining threshold alerts warns you before a disk fails.

How do you plan backups and disaster recovery on a dedicated server?

Backups on a dedicated server are entirely your responsibility, and at least one copy of the data must be kept off the server. RAID protects against drive failure; but against accidental deletion, software bugs, ransomware or a problem affecting the whole server, only an off-server backup helps.

• File and database backups: incremental, encrypted backups with restic, borg or Proxmox Backup Server.
• Virtual machine backups: if you use Proxmox VE, send scheduled VM backups to separate storage.
• Configuration: if you codify your server setup with a tool like Ansible, you can bring a new machine to the same state in minutes rather than hours.
• Recovery time objective: have a written answer to "if the server is completely lost, how many hours until we are back?" and test it at least once a year.

Is it better to rent or buy a dedicated server?

For most businesses, renting a dedicated server makes more sense than buying one and hosting it in the office or a data center, because hardware failures, spare parts, cooling, power, network connectivity and DDoS protection are included in the rental. With a purchased server, each of these is a separate cost and a separate responsibility.

Renting stands out when:

• You want enterprise-class hardware without tying up capital up front.
• You do not want downtime while waiting for parts after a hardware failure.
• You plan to move to a newer platform within a few years.

Buying may make sense for very long-term, unchanging projects that need special hardware; in that case placing the hardware in a data center (colocation) may be an option. If you are still deciding between virtual and physical, a VPS server is the low-cost entry point.