Back to Blog
2026-04-14Knowledge Tree Team9 min readCase Study

We Discovered 10,000 Resources Nobody Knew Existed

discoveryshadow-ITcase-study

Day 1: The Deployment

The email from the CTO was blunt: "We need to know what we own. All of it. By end of quarter."

The infrastructure team at a 3,000-person SaaS company was about to discover just how little they actually knew about their own infrastructure. They deployed Knowledge Tree on a Tuesday morning. By Tuesday afternoon, everything changed.

The setup took 45 minutes. They connected AWS (23 accounts), Azure (4 subscriptions), and 6 Kubernetes clusters. Read-only credentials — no risk, no modifications.

Then they clicked "Discover."

Day 1: The First Shock

Within an hour, the dashboard showed 47,000 resources. The team's best estimate had been 30,000. They were off by 17,000 resources.

Where were the extra ones?

  • 12 Development environments that no one remembered creating. Total cost: $43,000/month.
  • 847 S3 buckets — 312 of which hadn't been accessed in over 6 months
  • 156 EC2 instances running in regions the team didn't know they had enabled
  • 23 RDS databases with no corresponding application or owner tag

Week 1: The Deep Dive

Over the next week, the team used Knowledge Tree's knowledge graph to trace relationships and ownership. The discoveries kept coming:

Shadow IT

A marketing team had spun up their own CloudFormation stack — complete with a production-grade RDS database, ECS cluster, and CloudFront distribution. Total monthly cost: $12,000. No one in engineering knew it existed.

The Forgotten Migration

Two years ago, the company migrated from us-east-1 to us-west-2. Except they didn't finish. 340 resources were still running in the old region, connected to nothing, serving no traffic. Monthly cost: $28,000.

Security Gaps

The graph view revealed 23 security groups with port 22 (SSH) open to 0.0.0.0/0. Three of those were attached to databases containing customer data. The security team was notified within minutes.

Week 2: The Cleanup

Armed with complete visibility, the team began the cleanup:

  1. Decommissioned 4,200 orphaned resources — saving $180,000/year
  2. Consolidated 23 AWS accounts down to 15
  3. Remediated all 23 overly-permissive security groups
  4. Tagged every remaining resource with owner, team, and cost center

The knowledge graph made this safe: before decommissioning anything, the team could see exactly what depended on it. Zero production incidents from the cleanup.

Month 1: The Cultural Shift

Something unexpected happened. The team stopped guessing and started knowing.

  • Standup conversations changed from "does anyone know what this does?" to "the graph shows it connects to these 3 services"
  • Incident response accelerated because blast radius analysis took seconds instead of hours
  • Architecture reviews started with the topology map instead of a blank whiteboard
  • New hire onboarding went from 3 months to 3 weeks — the graph teaches new engineers the system

The Numbers

After 30 days with Knowledge Tree:

MetricBeforeAfter
Known resources~30K47K (100% coverage)
Monthly wasteUnknown$180K/year recovered
Security gapsUnknown23 critical findings remediated
Incident MTTR4.2 hours1.7 hours
Onboarding time3 months3 weeks

The Lesson

You can't manage what you can't see. And at scale, you're always seeing less than you think. The gap between what you believe you have and what you actually have grows by ~10% every quarter.

Knowledge Tree closes that gap. Not once, but continuously — because discovery isn't a one-time event, it's an ongoing practice.


What don't you know about your infrastructure? Start a free trial and find out in 30 minutes.