Private Kubernetes on bare metal, OpenShift and OKD

What it can do for you

Persistent storage for stateful workloads. Block volumes and shared read-write-many filesystems, replicated across nodes, so databases, queues and shared content survive the loss of a node. Volumes can be expanded in place, snapshotted and restored.

Real public IPs, no NAT, no extra appliance. Your services get public addresses announced over BGP directly from the cluster nodes to our network. No NAT in the path and no separate load balancer box in front of Kubernetes.

Failover in seconds. Every node advertises the service address, and fast link failure detection on the network side means traffic moves away from a failed node within a few seconds.

Isolation per tenant. Each tenant can get its own ingress controller, its own public IP, its own domain and an automatically renewing certificate, kept apart from other tenants and from the management network.

A bridge to what you already run. Containers that must reach your existing servers or databases do so through a dedicated egress identity on a private transit network, firewalled at the perimeter, so you can modernise one service at a time.

Databases built to survive maintenance. When we run clustered databases on the platform, such as Galera with Percona XtraDB Cluster, they are built with pod disruption budgets that keep quorum during rolling node maintenance, retained storage that is never deleted with a claim, and replicas spread across nodes.