Endpoint Reliability Studio
Five weeks to turn scattered API metrics into a shared reliability canvas your whole platform group can read.
What you will leave with
- A living route grid for one production service family
- Fragility scores tied to ownership, not vanity uptime
- A narrative review format for product and support partners
- Alert clustering rules you can defend in a postmortem
Instructor
Ana Vuković
Modules
Inventory without the spreadsheet fog
Capture endpoints, owners, and SLOs as nodes you can rearrange in a shared canvas.
Route Heat Overlay
Layer latency and error density onto the map so fragile edges glow before paging anyone.
Dependency storytelling
Trace how one slow authentication hop ripples through checkout and billing routes.
Review rituals
Run a 25-minute reliability walkthrough that product managers finish without glossary cards.
Noise diet clinic
Retire duplicate alerts and document which routes still deserve a wake-up call.
FAQ
Do I need a specific observability vendor?
No. Exercises use exported CSV and OpenTelemetry-friendly samples. You adapt overlays to whatever your company already pays for.
Is GraphQL covered?
Lightly. The studio focuses on REST and gateway routes. GraphQL resolver trees are mentioned, but you will not build a full GraphQL heat map in class—that remains a known limitation of the current curriculum.
How much live time each week?
One 90-minute studio session plus an optional 40-minute office hour. Async exercises take about three hours weekly.
Learner notes
Week three’s dependency exercise exposed a retry storm we had blamed on the CDN. Short and blunt: the map did the arguing for me.
★★★★☆ — “Clear visuals. Office hours sometimes filled up fast, so record the session if you join from a noisy timezone.”