Cut deployment failures by 80% across a 100+ cluster Kubernetes estate | Case Study | The Software Geek
All case studiesRegulated industries · Multi-cloud

Cut deployment failures by 80% across a 100+ cluster Kubernetes estate

Client A multi-sector investment groupServiceKubernetes & Platform EngineeringDuration 5+ years

An investment group ran a portfolio of operating companies, each on its own infrastructure and each answering to its own regulator. The job was to make that estate behave like one platform without flattening the separation the regulators required.

80%Fewer deployment failures
100+Clusters under one architecture
6 wks → 1 dayInfrastructure provisioning time

The challenge

  • Every operating company fell under a different regulatory regime, so one shared platform was never going to be permissible.
  • Workloads were split across AWS and GCP with no shared platform standards.
  • Provisioning new infrastructure took around six weeks of coordination.
  • Deployment failures were frequent enough to erode confidence in releasing.

What we did

  • Designed a multi-tenant Kubernetes architecture spanning EKS and GKE, eventually 100+ clusters.
  • Moved delivery to GitOps through ArgoCD so cluster state became reviewable and reversible.
  • Built self-service infrastructure on Terraform, Crossplane and Helm so teams provisioned without a ticket.
  • Established landing zones, account structure, IAM and identity federation, network segmentation and disaster recovery across both clouds.
  • Codified PCI DSS and GDPR obligations as policy-as-code governance rather than review checklists.

Technology

KubernetesEKSGKEArgoCDTerraformCrossplaneHelmIstioAWSGCP

Facing something similar?

Tell us what you are working on and we will come back with a straight answer on whether we can help, and what it would take.

Schedule Free Consultation

More work