RACK TRAIL
Solutions
VMware MigrationOpenStack at scaleProxmox at scalePrivate AICloud Cost OptimizationManaged Private CloudLarge Scale MigrationBig Data InfrastructureConfidential Computing InfrastructureCephKubernetes WorkloadsCloud Infrastructure for Startups
Why RacktrailGroundworkAbout Talk to an Engineer
Products · Bare Metal

Dedicated hardware. Nothing between you and it.

Single-tenant physical servers in the region you need. Run any hypervisor, any operating system, or nothing at all.

What you get

The whole machine.

No hypervisor tax, no shared tenancy, no neighbours competing for your IOPS.

Out-of-band management.

IPMI/Redfish access. Reboot, reinstall, and console in without opening a ticket.

Private networking.

Layer 2 between your machines, VLANs, and private interconnect to your private cloud if you run both.

Your choice of everything.

Proxmox, KVM, VMware, Hyper-V, bare Linux, or a Kubernetes node pool.

Configurations

Profile CPU RAM Storage Network Pricing
Compute AMD EPYC 9354 32c / 64t · 3.25 GHz 384 GB DDR5 ECC RDIMM 2 × 3.84 TB NVMe 1 DWPD, mirrored 2 × 25 GbE LACP Request a quote
Storage AMD EPYC 9224 24c / 48t · 2.50 GHz 256 GB DDR5 ECC RDIMM 12 × 20 TB SAS + 2 × 1.92 TB NVMe cache 2 × 25 GbE LACP Request a quote
Memory-optimised 2 × AMD EPYC 9534 128c / 256t 1.5 TB DDR5 ECC RDIMM 2 × 7.68 TB NVMe 3 DWPD, mirrored 2 × 100 GbE LACP Request a quote

Every configuration is quoted before you sign. Custom builds and multi-node fleets are scoped on a call.

What people run on it

Databases

PostgreSQL, MySQL and ClickHouse, where predictable I/O matters more than elasticity.

Analytics

Large datasets that get expensive fast on metered infrastructure.

Virtualisation hosts

Your own Proxmox or KVM estate, on hardware you specify.

Storage clusters

Ceph and MinIO, where you want the disks to be yours.

CI runners

Build fleets that would cost more per month than the hardware.

Latency-sensitive services

No hypervisor between your process and the NIC.

Bare metal or virtual machines?

Choose bare metal when

Your workload is steady rather than spiky, performance is predictable and measurable, you need hardware-level control, or virtualisation overhead is costing you real money.

Choose a private cloud when

You need to provision and destroy frequently, you want an API and self-service for your own teams, or your workloads are numerous and small.

Most teams run both. Databases and storage on metal, application tier on the cloud, one private network across them.

The data path

No hypervisor between your process and the NIC. What the hardware does for a given workload depends on the configuration — tell us what you need to run and we'll size it with you.

Data path · nothing between the process and the NIC Compute profile
NO HYPERVISOR nothing between you and it PROCESS 2 × 25 GbE 2 × 3.84 TB NVMe · mirrored 1.42M IOPS · 0.21 ms p99

Tell us what needs to run.

Send us the workload, the constraints, and the region it has to live in. An engineer — not a sales rep — will tell you honestly whether we're the right fit.