80% Cost Reduction — Non-Prod Migration from GCP to E2E Networks
Migrated the entire non-production environment for a Kochi-based AI recruitment startup from GCP to E2E Networks — cutting monthly cloud spend by 80%.
Context
The startup was running all non-prod environments on GCP at ₹1.5L/month. For workloads that didn’t need hyperscaler-grade infrastructure, this was an expensive setup. The goal was to find a cost-effective alternative without compromising developer experience or reliability.
What I Did
- Validated feasibility by deploying a single-node Kubernetes cluster on E2E using kubeadm and running performance benchmarks
- Configured a production-grade, secure Kubernetes environment on E2E Managed Kubernetes
- Deployed the full application and dependency stack — PostgreSQL cluster, Redis cluster, Elasticsearch (AI vector DB), HashiCorp Vault, Cert Manager
- Built the observability layer — Prometheus, Loki, Grafana
- Set up CI/CD with GitLab, self-hosted runners, and ArgoCD for GitOps deployments
- Migrated all non-prod environments from GCP to E2E and validated end-to-end
Results
| Metric | Before (GCP) | After (E2E) |
|---|---|---|
| Monthly spend | ₹1,50,000 | ₹25,000–30,000 |
| Cost reduction | — | ~80% |
| Delivery timeline | — | 1 month |
Stack
Kubernetes · E2E Managed K8s · kubeadm · PostgreSQL · Redis · Elasticsearch · HashiCorp Vault · Cert Manager · Prometheus · Loki · Grafana · GitLab CI · ArgoCD