Design a Real-time Analytics Dashboard

Design a real-time analytics dashboard for a global e-commerce marketplace that allows operations teams to monitor live business and system metrics. The dashboard must ingest 100 000+ events per second (orders, payments, page-views, errors, latency samples) and display continuously updating widgets such as revenue per minute, orders per second, payment-success rate, 99-th percentile latency, and top-selling items. Updates must appear on the dashboard within one second of the event being generated. The system should support 5 000 concurrent dashboard viewers, each able to customize the time range (last 5 min, 1 h, 24 h, 7 d) and granularity (1 min, 5 min, 1 h). Historical data older than seven days is not required in real time and may be served from a separate data warehouse. Design the end-to-end architecture: event ingestion, stream processing, storage, aggregation strategy, API layer, and the push mechanism that delivers live updates to browsers. Discuss how you would guarantee at-least-once ingestion, handle late-arriving events up to 30 s, scale individual components, and survive zone failures with minimal downtime.