Log ingest with PII redaction
Your task
Design the ingest-time processing stage of a log platform for a regulated fintech: 500GB/day of logs from 300 services in inconsistent formats (JSON, logfmt, raw text). Requirements: parse/normalize into a common schema, redact PII (card numbers, SSNs, emails) before it's ever persisted, drop noisy debug lines to control cost, and route security-relevant logs to a separate compliance store with 7-year retention while app logs keep 30 days. Design the processing pipeline and the routing.
It covers
- Parsing
- Schema on read
- Pii redaction