Confidential. The client and platform names are withheld under NDA. This study refers to the system as "the Platform."

Case Study · Enterprise Platform Engineering

An enterprise multi-cloud control plane

A governed control plane for regulated enterprises, with provisioning, FinOps, and audit built in from the first commit. Engineers now at Fortstak were among the founding team that designed and built it.

What it is
Enterprise multi-cloud management platform
Built for
Enterprises operating across multiple clouds and datacenters
Reach
Five public clouds plus a private datacenter on Apache CloudStack
Fortstak's role
Founding-team engineers who designed and operate it

01 · The ChallengeMulti-cloud without a control plane is risk on the balance sheet.

Enterprises rarely live on a single cloud. Workloads spread across AWS, Azure, GCP, and their own datacenters, each with its own console, billing model, and access controls.

The result is predictable: no single view of what is running, inconsistent governance between environments, cloud spend that drifts faster than anyone can track, and audit gaps that surface at the worst possible moment, when a regulator or a customer's security team asks. For any organization under real governance or compliance obligations, that is not an inconvenience. It is risk on the balance sheet. These teams need to provision quickly without losing control, and to keep control without slowing delivery to a crawl.

02 · The SolutionOne governed marketplace across every cloud.

Engineers now at Fortstak were among the founding team that designed and built the Platform, a unified control plane that turns several clouds into one governed marketplace. Teams request resources (virtual machines, databases, Kubernetes clusters, networks) from a single interface. The Platform provisions them across any supported target, tracks cost per project and department, enforces quotas and policy, and records every action for audit.

It spans both worlds: the five major public clouds and a private datacenter running Apache CloudStack, provisioned through the same governed workflow. Onboarding is deliberately approval-based rather than open self-signup: every organization is validated before it touches a single cloud resource, matching the expectations of enterprise-grade environments.

03 · ArchitectureRequest to provision, end to end.

The design is built around four principles regulated enterprises actually demand: asynchronous reliability, isolated execution, least-privilege security, and complete auditability.

FORTSTAK · ENTERPRISE PLATFORM ENGINEERING Multi-Cloud Control Plane Architecture One governed control plane across five public clouds and a private datacenter (Apache CloudStack). SECURITY HashiCorp Vault secrets store Dynamic secrets short-lived creds Least privilege scoped access Ephemeral exec destroyed per job Full audit every access logged OBSERVABILITY Prometheus metrics Grafana dashboards & SLOs Loki · Tempo logs & traces Zabbix host & service Full coverage every resource Enterprise User requests VMs, databases, clusters, networks 1 Marketplace Portal Next.js · next-auth Catalog, requests, dashboards, invoices 2 RabbitMQ Durable queue · retries · horizontal scale 3 CONTROL PLANE Engine API Node.js · Express · Prisma Validates quotas & policy, orchestrates every job PostgreSQL business & billing Neo4j resource topology MongoDB audit trail Redis cache · limits 4 Ephemeral Automation Pods Kubernetes (AKS · EKS · GKE) · Ansible Spawned per job, fetch secrets, then destroyed secrets 5 Public clouds AWS · Azure · GCP · OCI · Alibaba Private DC Apache CloudStack 6 Provisioned & governed resources telemetry 7 Notifications Email · SMS · Push, on completion or failure Asynchronous by design · isolated, ephemeral execution · least-privilege throughout · complete audit trail © Fortstak · reference architecture · platform name withheld

Reference architecture. Every request flows through a durable queue, a policy-checking control plane, and isolated ephemeral execution before touching any cloud or the private datacenter.

04 · How It WorksEngineered for what an auditor and a CFO care about.

Asynchronous by design

Every request is published to a durable message queue, so the platform stays responsive, scales horizontally under load, and retries failed jobs instead of losing them.

Isolated, ephemeral execution

A central engine validates quotas and policy, then spawns short-lived automation pods on multi-cloud Kubernetes (AKS, EKS, GKE) that run one job and are destroyed. Nothing runs with standing credentials.

Least-privilege security

Every execution pulls scoped, short-lived secrets from a vault. Access is documented and time-bound, and every secret retrieval is logged.

Complete auditability

State is separated by purpose: relational for billing, a graph for live topology and dependencies, a document store for the full audit trail. Who did what, where, and when is always answerable.

05 · What It DeliversNot a dashboard. A control plane.

Unified control

One console across public clouds and a private datacenter, no per-vendor context switching.

Governance and tenancy

Users, departments, projects, roles, and permissions, with quotas and policy enforced everywhere.

FinOps

Daily cost tracking, invoices per project and provider, and detection of orphaned resources.

Compliance and audit

Approval-based onboarding and comprehensive audit trails mapped to regulatory expectations.

Security by default

Vault-managed credentials, least-privilege access, and isolated, ephemeral execution.

End-to-end observability

Metrics, logs, and traces across every provisioned resource.

06 · Why It Matters

Most multi-cloud tools show you the sprawl. The Platform governs it.

It was built for environments where a standing credential or a missing audit trail is a finding, not a footnote, and where provisioning speed and control are expected to coexist rather than compete.

07 · TechnologiesThe stack behind it.

Kubernetes (AKS · EKS · GKE)Apache CloudStackDockerAnsibleRabbitMQHashiCorp VaultPostgreSQLNeo4jMongoDBRedisPrometheusGrafanaLokiTempoZabbixNext.jsNode.js

08 · What This DemonstratesThe depth our engineers bring.

The Platform is evidence of how our people work. They were among the founding team that designed and operated an enterprise-grade, multi-cloud system where security, governance, and cost control were built in from the first commit, not bolted on before an audit. That is the standard they now bring to Fortstak engagements.

Running across more than one cloud?

If your teams are losing visibility, governance, or budget in the gaps between clouds, or between cloud and your own datacenter, that is precisely the problem we build for. The first step is a fixed-scope platform review.

Talk to Fortstak →