A FinOps-Integrated Platform Engineering Framework for Cost- Aware Cloud Resource Provisioning and Optimisation

Abstract

Rising cloud adoption has increased enterprise access to scalable computing, but it has also intensified uncontrolled expenditure, overprovisioning and
weak cost ownership. Conventional platform-engineering practices improve delivery speed and developer self-service, yet often apply financial controls
after deployment. This study designed and evaluated a FinOps-integrated platform-engineering framework that embeds cost governance within resource
provisioning and continuous optimisation. A design-science method was combined with a reproducible AWS Elastic Kubernetes Service simulation
comprising 30 paired runs, 40 services per run and 30 days of workload at five-minute resolution. The framework integrates an internal developer platform,
pre-deployment cost estimation, budget guardrails, Infrastructure as Code templates, rightsizing, workload forecasting, automated policy enforcement and
continuous cost feedback. Compared with conventional provisioning, the framework reduced mean monthly infrastructure expenditure from $2,840.53
to $1,896.08, representing a 33.25% reduction. Mean CPU utilisation increased from 23.85% to 36.20%, a gain of 12.35 percentage points, while policy
violations declined from 63.60 to 9.67 per 160 provisioning requests, an 84.80% reduction. Mean provisioning time increased from 7.00 to 7.20 minutes,
or 2.82%, remaining within the predefined ±5% equivalence margin. Simulated availability changed from 99.9957% to 99.9808%, remaining within the
0.10-percentage-point noninferiority margin. The study contributes a preventive cost-governance model, empirical simulation evidence and practical guidance
for implementing financially accountable cloud platforms without materially reducing provisioning efficiency or reliability.