Virtual machines, containers, storage and networking on servers you already own, or in the cloud you already use. Every site, every node and every workload, from one portal.
No separate hypervisor, storage array, container platform and network tool to buy, license and patch. Kantrix runs all of it on the same nodes.
Start from a built-in image or your own ISO. VMs boot with UEFI, Secure Boot and a virtual TPM, and Windows installs need no extra drivers.
Deploy ready-made apps from Hub, or your own containers and apps, to any node, on the same hardware as your VMs.
In a cluster, each VM's disk is copied to a second node by default (RF2). Choose three copies on three nodes (RF3) for the workloads that matter most, so a failed server doesn't take its data with it.
Nodes connect out over encrypted links. You don't open inbound ports or put anything on the public internet.
Authenticator apps and passkeys for every user, roles for each person, and an audit trail of every change that no one can edit or delete, not even us.
Assign a node's GPU directly to a VM, so AI workloads run on hardware you own. How it works ↓
Describe it in plain words in the portal: a file server for the new office, a test copy of the accounting VM, room for a new app. Kantrix works out the VMs, storage and networking, and builds it on your nodes.
Pass a node's GPU straight through to a VM. The VM gets the card itself, with its own driver and its full memory, so training and inference run at the speed of the hardware. Your data stays on servers you own, at the site you choose.
Needs a server with IOMMU turned on in its firmware (Intel VT-d or AMD-Vi). Your partner can check yours.
Boot a server from the Kantrix installer, or start a VM in your cloud. One install, no separate hypervisor.
Scan the code on the node's screen with your phone, and it joins your fleet at the right site.
Create VMs, deploy apps and add nodes to a cluster from the portal, for one site or a hundred.
Nodes need wired Ethernet with DHCP and outbound internet access.
Most businesses don't have a server room, a storage team or a spare rack. Kantrix is built for two servers in a closet as much as for twenty.
A two-node cluster with a witness keeps a copy of every disk on each server. No third box to buy.
VMs on replicated storage keep running if the site loses its internet link.
Keep the backup product you trust inside each VM. Kantrix works alongside it.
A server in the back room and a VM in a public cloud are both nodes in one fleet. No hardware list to buy from, and no capacity to license.
Any x86 server with wired Ethernet. For GPU passthrough, the server also needs IOMMU turned on in its firmware.
Two nodes are free. Beyond that, your partner quotes you. The MSRP is $150 USD per node, per month, for the whole platform, on any hardware or in any cloud.
No. A node is a node, whatever is in it.
Yes. Run its agent inside each VM, as you do today.
Your workloads and their data stay on your nodes. Read the security overview.
A partner near you can run Kantrix on one of your servers and show you.