Building My First FinOps Dashboard

So there I was last Tuesday, staring at our GCP bill that somehow doubled since last month, when my manager walks by and casually drops: "Hey, we need a FinOps dashboard by Friday."

The Beginning of My FinOps Nightmare (Er... Journey)

Let me tell you what they don't mention in those fancy cloud certification courses - nobody warns you about the existential crisis you'll have trying to make sense of billing data across three cloud providers.

First thing I learned: our "tagging strategy" was basically "let the intern decide" for the past year. Cool, cool.

What I Actually Built (After 27 Coffee Cups)

1. Cost Visibility Panel (AKA "The Panic Button")

• Created a real-time spend tracker that practically screams when we exceed budgets

• Built departmental breakdowns that immediately started a blame game between Dev and QA

• The service breakdown chart revealed we're spending more on idle GKE clusters than production (oops)

2. Resource Efficiency Section (Or "The Wall of Shame")

• Found 7 Kubernetes clusters nobody would claim ownership of - they've been running for 9 months

• Discovered our usage patterns spike every Thursday because someone scheduled automated pods to run infinitely

• The best part? Our ML team was running n1-standard-32 nodes to host documentation websites

3. Optimization Insights (The "How We Keep Our Jobs" Tab)

• Committed use discount recommendations showing we could save 34% if we planned anything longer than 24 hours ahead

• Built an automated Slack alert that tags the engineer who spun up that 96-core GKE node pool "just to test something"

• Created an anomaly detector that caught a rogue Kubernetes deployment that somehow kept scaling itself to 200 replicas

What Actually Happened When I Presented It

Marketing: "Can we make the graphs prettier?" Finance: "So you're saying we've been wasting HOW much?" Engineering: "I'm pretty sure those 18 Horizontal Pod Autoscalers are mission-critical" (Narrator: They weren't)

The real MVP feature? A simple button labeled "Who's responsible?" that maps namespace resources to teams. Suddenly everyone's very interested in optimization.

The Brutal Truth

Building this dashboard was like performing surgery while learning anatomy. I spent hours digging through BigQuery exports trying to understand why our Cloud Run services cost more than our entire infrastructure last year. But now that it's running, I can finally sleep at night knowing that random StatefulSet won't bankrupt us.

Next challenge: convincing people to actually USE the dashboard. Thinking of installing monitors in the bathroom stalls - they can't avoid it there!