Computational backend

UpCloud is built on the principles of speed and reliability by employing the latest technologies and hardware.

Compute hosts

UpCloud’s infrastructure is designed from the ground up with high availability in mind. Cloud Servers are run on a compute hosts, on which all mission-critical components are backed up by N+1 redundancy, including dual network connections and error-corrected memory modules.

Block storage is provided from an equally highly available and fault tolerant storage system.

Automated failover

Should a compute host fail, all its Cloud Servers will be transferred over to another compute host and the Cloud Servers will continue running with minimal interruption. A live migration will be attempted whenever possible, causing only an interruption of a few seconds to the server’s operation.

Virtualisation

We use the true and tested Linux Kernel-based Virtual Machine, or KVM, virtualisation module to provide hardware-assisted virtualisation.

CPUs

We use top-of-the-range hardware at every one of our data centres, including high-powered enterprise-grade processors. The exact make and model of compute hardware varies by generation. The newest hardware configurations use the AMD EPYC family of processors.

Automated balancing

UpCloud’s infrastructure optimises the load across compute hosts within data centres by deploying new and recently started Cloud Servers on the least used hosts. The compute host is reselected each time the server is started. Users experiencing reduced performance can take advantage of the automated balancing by shutting down and starting the affected Cloud Server.