Kubernetes platform implementation
Production-ready Kubernetes and OpenShift clusters on bare metal, VMware, OpenStack or the cloud, with the platform services around them, built as code and tested before you go live.
A platform built for production from the first day
A cluster that runs a demo is not a platform. Production needs high availability, networking that fits your data centre, storage for stateful workloads, single sign-on, a trusted registry, monitoring, logging, backups and security controls, all working together and documented.
We build all of it as code, so every cluster is consistent and can be rebuilt. Before handover we test it the way production will use it: failing nodes, restoring backups, upgrading to a new version and running your pilot application under load.
When it fits
01
The platform is chosen
You know which Kubernetes you want, and need it built properly in your data centre or cloud.
02
Hardware is arriving
Servers, storage and network equipment are bought or on order, and need to become a platform.
03
Pilot clusters need replacing
Early clusters were built by hand, and production should be rebuilt cleanly, as code.
04
A new site is planned
A disaster recovery site, a second data centre or a branch location needs a cluster that matches the first.
What we do
Cluster build
Highly available control planes and worker pools on bare metal, VMware, OpenStack or the cloud, installed by automation.
Networking and traffic
Container networking, ingress and load balancing, DNS and network policies, connected to your firewalls and existing networks.
Storage for stateful workloads
Storage classes on your SAN, NAS or software-defined storage, with snapshots and the performance databases need.
Identity and access
Single sign-on with Active Directory or LDAP, access by team and role, and audit logging.
Registry, monitoring and logging
A trusted image registry, metrics, dashboards, alerts and central logs, ready before the first application arrives.
Hardening and acceptance tests
Hardening against the CIS benchmark, then failure, backup and restore, upgrade and load tests with a pilot application.
What we build
Five layers, built and tested together
Every layer is defined in Git and applied by automation, so the next cluster, site or rebuild comes out the same.
- Infrastructure
- Bare-metal servers, VMware, OpenStack or cloud, with the network and storage underneath.
- Kubernetes
- Highly available control planes and worker pools, on a supported version.
- Platform services
- Networking, storage, registry, identity, monitoring, logging and backup.
- Delivery
- GitOps and pipelines that deploy the platform and the applications on it.
- Applications
- Your workloads, starting with a pilot that proves the platform.
How we build a platform
01
Prepare
Confirm the design, check that hardware, networks and firewalls are ready, and set up the automation and Git repositories.
02
Build
Install the clusters and platform services as code, connect identity, storage and monitoring, and harden the configuration.
03
Test
Run failure, restore, upgrade, security and load tests, and fix what they find.
04
Go live and hand over
Onboard the pilot application, train your team, and hand over the documentation and runbooks.
What you get
- Production clusters, built and configured as code
- Ingress, storage, registry, identity, monitoring, logging and backup, working together
- A hardening report against the CIS benchmark
- Results of failover, restore, upgrade and load tests
- Architecture documents and runbooks
- Training and handover for your platform team
Platforms and tools
- Red Hat OpenShift
- Kubernetes
- Amazon EKS
- Azure Kubernetes Service
- Google Kubernetes Engine
- Terraform
- Ansible
- Argo CD
- Cilium
- Calico
- MetalLB
- Harbor and Quay
- Prometheus and Grafana
- Loki
- Velero
- Keycloak
Questions about implementation
Can you build on our existing VMware or OpenStack?
Yes. Kubernetes and OpenShift run on vSphere and OpenStack and use their networking and storage. Bare metal gives the best performance and avoids hypervisor licences, so we compare both with you.
OpenStack platformsHow many clusters do we need?
Usually fewer than teams expect. Most organisations start with a production and a non-production cluster at each site, and separate teams with namespaces, quotas and policies rather than more clusters.
Can you install without internet access?
Yes. For closed networks we mirror the images and updates into a registry inside your network and install from there, which many government and financial clients require.
Will our team be able to run it?
That is part of the build. Your engineers work alongside ours, every change is in Git, and handover includes runbooks and training. We can also run the platform for you.
Operations and supportBuild a platform you can rely on
Tell us which platform you have chosen and where it will run. We'll plan the build, the tests and the handover.