home / case studies / US-headquartered logistics software firm with India engineering centre (composite)
DevOps Platform

Multi-Cloud Kubernetes Consolidation, Observability and FinOps

We consolidated sprawling clusters and VMs into a standard Kubernetes platform with unified observability and cost visibility. The client cut idle spend and gained a single view of reliability across clouds.

US-headquartered logistics software firm with India engineering centre (composite) · DevOps Platform

Multi-Cloud Kubernetes Consolidation, Observability and FinOps

Challenge

Migrating live workloads without downtime and changing team habits around resource requests. We used phased namespace migrations, automated request recommendations and SLOs agreed with each product team.

Approach

We standardised on a cluster blueprint (Terraform + Cluster API) for AWS and Azure, with Karpenter/cluster-autoscaler, shared ingress, cert management and secrets via External Secrets and a vault. Observability uses OpenTelemetry, Prometheus, Loki and Tempo with Grafana dashboards and SLO-based alerts. Kubecost-style cost allocation by team and namespace feeds monthly showback reports. Right-sizing and spot/preemptible node pools were introduced for stateless workloads.

Outcome

Illustratively ~25–40% lower compute spend through right-sizing and autoscaling Fewer, more actionable alerts tied to SLOs Team-level cost showback driving better decisions One blueprint for new clusters across clouds Faster incident triage with correlated logs, metrics and traces

~25–40%Illustratively lower compute spend through right-sizing and autoscaling
✓Fewer, more actionable alerts tied to SLOs
✓Team-level cost showback driving better decisions

Problem solved

Dozens of inconsistently configured clusters, alert fatigue and no clear link between cloud bills and teams.

What we built

We standardised on a cluster blueprint (Terraform + Cluster API) for AWS and Azure, with Karpenter/cluster-autoscaler, shared ingress, cert management and secrets via External Secrets and a vault. Observability uses OpenTelemetry, Prometheus, Loki and Tempo with Grafana dashboards and SLO-based alerts. Kubecost-style cost allocation by team and namespace feeds monthly showback reports. Right-sizing and spot/preemptible node pools were introduced for stateless workloads.

Benefits

Illustratively ~25–40% lower compute spend through right-sizing and autoscaling Fewer, more actionable alerts tied to SLOs Team-level cost showback driving better decisions One blueprint for new clusters across clouds Faster incident triage with correlated logs, metrics and traces

KubernetesTerraformCluster APIOpenTelemetryPrometheusLokiTempoGrafana

Talk to us