hard🏗️ System Design

Design a real-time supply chain visibility platform for a Fortune 500.

A Fortune 500 retailer wants a platform that gives their ops team live visibility into inventory levels, shipment status, and supplier delays — across 500 warehouses globally. **Requirements:** - < 5 min data freshness for shipment events - 10,000 concurrent users (ops analysts) - Support for 50+ ERP and WMS integrations - Alerting when KPIs breach thresholds **Design the system.**

💡 Hints (3)

1.Start with data ingestion — how do you pull from 50+ heterogeneous sources?
2.Think about the read vs write path separately.
3.Consider eventual consistency trade-offs for real-time vs batch.

✅ View Solution

**High-level architecture:** 1. **Ingestion layer** — Kafka + connector framework (Kafka Connect or custom adapters per ERP). Events normalized into a canonical schema. 2. **Stream processing** — Flink or Spark Streaming for real-time aggregation. 3. **Storage** — TimescaleDB or ClickHouse for time-series analytics; Redis for hot KPI snapshots. 4. **API layer** — GraphQL subscriptions for live dashboard updates. 5. **Alerting** — Rules engine triggers on threshold breach → PagerDuty / Slack. **Trade-offs to discuss:** pull vs push for ERP integration, cost of Kafka at scale, latency vs consistency.