Skip to content

cost_analysis.md

Latest commit

 

History

History
133 lines (83 loc) · 2.98 KB

cost_analysis.md

File metadata and controls

133 lines (83 loc) · 2.98 KB
Raw

Cost Analysis and Savings Modeling

Economic Validation of Cloud Optimization Engine

1️. Experiment Overview

This project was validated using real infrastructure deployed via Terraform in a clean Google Cloud project.

Two controlled states were measured:

  • Phase A (Baseline) – 1 running VM, 3 unattached disks, 2 reserved static IPs
  • Phase B (Scaled) – 1 running VM, 5 stopped VMs, 3 unattached disks, 2 reserved static IPs

All costs shown below are taken from Billing → Reports → Usage cost ($). Subtotal is zero because trial credits covered expenses, but usage cost reflects real infrastructure consumption. 1 credit = 1 CAD in this project, but the cost analysis is based on actual GCP pricing benchmarks.

2️. Baseline State (March 5–12)

Infrastructure

  • 1 × e2-medium (RUNNING)
  • 3 × unattached disks (200GB, 100GB, 50GB)
  • 2 × reserved static IP
  • 1 × stopped VM

Usage Cost (7 days)

SKU Cost (CAD)
Static IP Charge 4.84
Storage PD Capacity 3.06
Compute ~0
Total 7.91

Projection

Monthly ≈ 7.91 × 4 = 31.6 CAD
Annual ≈ 31.6 × 12 = ~379 CAD

3️. Scaled State (+4 Stopped VMs)

Additional stopped VMs each had 10GB boot disk.

Updated Usage Cost

SKU Cost (CAD)
Static IP Charge 5.43
Storage PD Capacity 3.69
E2 Instance Core (running VM) 0.41
E2 Instance RAM (running VM) 0.22
Total 9.84

Delta (1 day growth)

Increase from 7.91 → 9.84

Δ = +1.93 CAD

4️. Financial Projection After Scaling

Daily exposure ≈ 1.93 CAD increase trend

Projected monthly impact:

1.93 × 30 ≈ 57.9 CAD/month

Projected annual impact:

57.9 × 12 ≈ ~695 CAD/year

5️. Key Findings

🔴 Static IP Waste

Unused reserved IP addresses were the largest contributor to waste.

🟠 Storage Waste

Unattached disks and boot disks of stopped VMs continuously accumulate cost.

🔵 Compute Cost

Running compute becomes visible once free-tier coverage is exceeded.

6️. FinOps Insight

Stopped VMs do not generate CPU cost, but they:

  • Continue accumulating disk storage cost
  • Represent lifecycle mismanagement
  • Increase financial exposure when scaled

This experiment confirms:

  • Infrastructure scale directly impacts cost
  • Storage waste compounds silently
  • Static IP waste is often overlooked
  • Automated detection provides measurable financial value

7️. Strategic Conclusion

Even a small demo project generated:

~700 CAD/year projected exposure when scaled.

Extrapolated across:

  • Multiple environments
  • Multiple teams
  • Organization-wide deployments

The impact becomes significant.

Cloud Optimization Engine provides:

  • Continuous detection
  • Automated reporting
  • Financial visibility
  • Governance enforcement foundation

This validates both technical automation and economic reasoning in a real cloud environment.