💰 Business Case & CFO Approval

What happens when you click

🔎 Scan CMDB — calls GET api/servers?zombie=true; the backend scans all 5,000 server records and returns those with <1% average CPU over 90 days. The count becomes your “zombies identified” number.
🪦 Mark zombies for retirement — calls POST api/business-case/retire-zombies; server status changes to retired with strategy retire. These servers are excluded from the cloud cost forever (and from every later module).
Sliders — recalculate TCO client-side preview and then POST api/business-case/tco for the authoritative 5-year CapEx vs OpEx curve, charted with Chart.js.
📄 Generate Business Case PDF — builds a simulated document (text) from your current numbers and sets has_pdf = true.
🧑‍💼 Submit to CFO — POST api/business-case/cfo; the AI CFO rejects until zombies are retired, the PDF exists and savings ≥ 15%.

Step 1 — Scan the on-prem CMDB

Not scanned yet.

IDName / OSHypervisorvCPUCPU%

Step 2 — TCO sliders: on-prem CapEx vs cloud OpEx

Years modelled 5 Hardware refresh per server ($) 6000 Power & cooling per server / yr ($) 900 Real estate / yr ($M) 2.5 Staff cost / yr ($M) 6 Cloud $/server-hour 0.18 Reserved / Savings Plan discount % 35

Step 3 — Document & approval

No document yet. Generate the business case after calculating TCO.

BSAscan & model

➜

CMDB5,000 servers

➜

Zombiesretire

➜

TCO modelCapEx vs OpEx

➜

CFOapprove?

Click a node to learn what it does.

CFO verdict appears here.

Deep dive

How

Pull 90-day CPU/memory telemetry per server from the CMDB or discovery tool.
Model 5 years: on-prem has CapEx spikes (refresh), cloud is linear OpEx.
Exclude zombies, apply reserved-instance discounts, add FinOps staff.

Why

CFOs sign business cases, not architectures.
A migration that moves waste just rents waste at a higher hourly rate.
CapEx → OpEx changes balance-sheet treatment and cash-flow timing.

When

T-12 to T-9 months, before mobilising the 25-person team.
Refresh the case each SteerCo as real cloud bills arrive.

Pitfalls

Comparing on-prem list prices to on-demand cloud prices (ignores commitments).
Forgetting egress, Oracle/Windows licensing and dual-running costs during the migration.
Claiming savings from day one — the first year is usually more expensive.

Key ideas

Assumptions matter more than the answer

The CFO will attack your assumptions, not your arithmetic. Document them explicitly (the API returns an assumptions list for this reason).

Trade-off

Reserved instances raise savings but lock you in for 1–3 years — fine for Rehost workloads, risky for apps you plan to refactor.

TCOCapEx vs OpExZombie serversHurdle rateFinOps

🐍 Python executed on the server (tco, cfo_review, retire_zombies)

loading…

loading…

loading…