How do you design a real-time database and historian architecture for large-scale SCADA?

Large-scale historian architecture: distributed collection (edge historians at remote sites, forwarding to central), tiered storage (high-resolution short-term, compressed long-term), replication for availability, and archiving strategy. Sizing: calculate tag count x collection rate x retention period, plan for growth. Performance optimization: efficient compression algorithms, indexed queries, calculated tags for aggregations. Integration: OPC connections to SCADA, PI-to-PI/store-forward for unreliable links, web APIs for application access. Consider: time synchronization (GPS, NTP), data integrity verification, disaster recovery, and regulatory compliance (21 CFR Part 11 for pharma). Implement monitoring and capacity planning.