Virtual Machines - Hikube

Reliable VMs,
100% hosted in Switzerland.

High-availability infrastructure in three Swiss data centers. Automatic restart in case of failure, replicated storage, data under Swiss jurisdiction. Deploy in minutes.

3 DC

Independent Swiss data centers

<5 min

Automatic restart after failure

64 vCPU

Maximum per VM

256 RAM

Maximum RAM per VM
Why Hikube

A sovereign infrastructure

For Swiss and European companies, hosting their data with an American hyperscaler exposes them to real jurisdictional risks. Hikube offers a high-performance alternative, without compromising sovereignty.

A 100% Swiss cloud - why it matters

In the face of the US Cloud Act and RGPD and nLPD compliance requirements, hosting your VMs in Switzerland is no longer a luxury, it's a necessity for companies handling sensitive data. Your data remains under Swiss jurisdiction, one of the most protective in the world.

Triple replication on 3 independent sites (Gland, Lucerne, Geneva) guarantees continuity of service even in the event of failure of a complete datacenter, without any manual intervention on your part.

A credible alternative to AWS, Azure and GCP for workloads requiring data sovereignty, regulatory compliance or independence from US suppliers.

Frame 11509
What you deploy on Hikube
markup-cropped

Migration from VMware or Proxmox

Import of custom VMDK, QCOW2 and ISO images. Transition without application rewriting.

markup-cropped

AI workloads with dedicated GPU

Attach an L40S, A100 or H100 GPU directly to a VM via PCI Passthrough, guaranteeing exclusive access to hardware and native performance.

markup-cropped

Legacy applications and specific OS

Windows Server, Linux distributions with custom kernel, non-containerizable software - deploy the exact OS and dependencies you need.

markup-cropped

Isolated and regulated environments

Finance, healthcare, public sector. Your VMs run in an isolated location, on an infrastructure whose geographical location you know exactly.

Available configurations

Three ranges for every load profile

The vCPU/RAM ratio determines the right range for your workload. Each range is available from small to 8xlarge.

S series

Standard - ratio 1:2

CPU-intensive workloads with low memory requirements.

s1.small 1 vCPU - 2 GB
s1.medium 2 vCPU - 4 GB
s1.large 4 vCPU - 8 GB
s1.xlarge 8 vCPU - 16 GB
s1.2xlarge 16 vCPU - 32 GB
s1.4xwide 32 vCPU - 64 GB

U series

Universal- ratio 1:4

CPU-intensive workloads with low memory requirements.

u1.medium 1 vCPU - 4 GB
u1.large 2 vCPU - 8 GB
u1.xlarge 4 vCPU - 16 GB
u1.2xwide 8 vCPU - 32 GB
u1.4xwide 16 vCPU - 64 GB
s1.8xwide 32 vCPU - 128GB

M series

Memory - ratio 1:8

High-memory instances for RAM-hungry applications.

m1.large 2 vCPU - 16 GB
m1.xlarge 4 vCPU - 32 GB
m1.2xwide 8 vCPU - 64 GB
m1.4xwide 16 vCPU - 128 GB
m1.8xwide 32 vCPU - 256 GB

Quick guide

Supercomputing → s1 - General applications → u1 - Databases and analytics → m1
Network & Access

Connect as you wish

Each VM is assigned a private IP in an isolated network. External exposure is optional and can be configured in two modes to suit your level of security.

PortList

Automatic firewall - dedicated IP

Only the ports you list are exposed. The firewall is applied automatically at service level, for a minimal attack surface without additional configuration.

WholeIP

Dedicated public IP - All ports

A public IP is assigned directly to the VM, and all ports are accessible. You need to configure a firewall in the VM itself.


Three ways to access your VM once deployed:


  • SSH
    Via virtctl ssh or directly via the external IP with your usual SSH client.

  • Serial console
    Low-level access directly to the VM, independent of network status. Useful if SSH connectivity is lost.

  • VNC
    Graphical interface via secure tunnel. Particularly useful for Windows VMs or desktop environments.

WholeIP - mandatory firewall in the VM

In this mode, all ports are open to the Internet. Configure ufw, iptables or nftables in the VM to protect your services.
starting up

Deploy in minutes

Two resources to create: a system disk (VMDisk) from an Ubuntu, Debian, Rocky or other image, and a VM (VMInstance) that references this disk, defines the configuration and injects your SSH key.

The cloudInit field lets you install packages, create users and run scripts on first boot, without manual login.

With storageClass: replicated, your disk is replicated across all 3 datacenters as soon as it is created.

yaml

apiVersion: apps.cozystack.io/v1alpha1
kind: VMInstance
metadata:
name: vm-instance
spec:
instancelype: u1.xlarge # 4 vCPU, 16 GB
instanceProfile: ubuntu
running: true
external: true
externalMethod: PortList
externalPorts:
- 22
- 443
disks:
- mon-disque
sshKeys:
- "ssh-ed25519 AAAA... user@host"
cloudInit: |
#cloud-config
packages:
- nginx
- htop
runcmd:
- systemctl enable nginx

AVAILABLE IMAGES

Supported operating systems

Pre-configured and optimized images for every distribution. Specify the image in your VMDisk or import a custom image from an HTTP URL.

UBUNTU

22.04 - 24.04

DEBIAN

12 - 13

ROCKY LINUX

8 - 9 - 10

ALMALINUX

8 - 9

CENTOS STREAM

9 - 10

ID format

ubuntu-2404, debian-12, rocky-9, almalinux-9. For VMware (VMDK), Proxmox (QCOW2) or custom ISO images, use the VMDisk manifest.
FAQ

Questions about Cloud Instances

What happens if the server hosting my VM goes down?

The platform detects the failure and automatically restarts your VM on another available server, in the same or another datacenter. Downtime is limited to restarting the VM, typically less than 5 minutes. Your data remains intact thanks to disk replication across multiple physical nodes.

Which instance series to choose?

The S series (1 vCPU for 2 GB RAM) is suitable for CPU-intensive workloads with little memory: CI/CD, compilation, computation. The U series (1:4) is the right starting point for most applications: web, APIs, lightweight databases. The M series (1:8) is designed for RAM-intensive workloads: SAP HANA, analytics, large Redis caches, data warehouses.

What's the difference between PortList and WholeIP?

PortList exposes only the ports you list, with an automatic firewall. This is the recommended mode for production - you retain exact control over what is accessible from the outside.

WholeIP assigns a public IP directly to the VM with all ports open. No automatic firewall - you have to configure one in the VM. Useful for VMs that act as gateways or VPNs, or for development where you need full access.

Can I migrate my VMware or Proxmox VMs to Hikube?

Yes, Hikube supports the import of VMDK (VMware), QCOW2 (Proxmox/OpenStack) and custom ISO images. Your existing VMs migrate without application rewriting.

How do I customize my VM on first startup?

Use the cloudInit field in the VMInstance manifest. Cloud-init runs on first boot and allows you to install packages, create users, inject SSH keys and execute shell commands. No manual connection is required for initial configuration.

Is my data really in Switzerland?

Yes, completely. The Hikube infrastructure is hosted in 3 datacenters in Switzerland (Gland, Lucerne, Geneva). Your data never leaves Swiss territory. Native RGPD and nLPD compliance. We can provide a DPA and answer your compliance questionnaires on request.