Interactive demos
Each demo solves a real problem. Failed payments, duplicate charges, cache stampedes, GDPR compliance. Click a button and it runs against the live cluster (Fly.io, Postgres, RabbitMQ). Nothing is mocked.
Live service status
Every checkout either completes fully or rolls back cleanly.
Zero lost events, even when the message broker goes down mid-write.
No silent overwrites. Conflicting edits are caught and surfaced immediately.
One failing service cannot take down the whole platform.
Duplicate clicks and network retries never double-charge a customer.
Abusive traffic is shed cleanly while legitimate users keep their quota.
Cache expiration does not flood the database. Only one caller rebuilds.
All servers converge within milliseconds of a write. No stale data served.
Failed refunds escalate to ops review. Never silently dropped.
Every transaction balances to zero. Money cannot be created or destroyed.
Data deletion touches all services or escalates. No partial deletes.
Search indexes stay current within seconds. No polling, no stale results.
No static passwords. Credentials rotate automatically without interrupting requests.
Safe retries
Click "pay" twice? Network retry? You are only charged once.
Duplicate clicks and network retries never double-charge a customer.
The Problem
Network retries create duplicate charges.
The Solution
Each request has a unique key. Duplicates return the cached result.
Sending the same request twice produces the same result. A duplicate click or network retry does not create a duplicate order.
- Click "Send request" to create an order
- Click it again with the same idempotency key
- The second request returns the existing order, not a new one
Safe retries
Advanced: Cache TTL Configuration
Lower TTL → replays after expiry produce a NEW order. Higher TTL → replays return the cached one.
Audit log
| Timestamp | Action | State |
|---|---|---|
| Fire a request from the controls above. this log will populate in real-time. | ||
Strategy: Postgres UNIQUE Constraint · TTL: 30s