Build a near real-time inventory dashboard with historical balances and thousands of updates per hour. Maximize dashboard performance and ensure accuracy. Which design do you implement?

The correct design leverages BigQuery's streaming inserts for near real-time updates. Streaming changes into a daily movement table allows for immediate capture of inventory transactions without directly modifying a large historical table. Calculating balances in a view that joins this movement table with a historical balance table provides up-to-the-minute inventory figures for the dashboard. A nightly update of the main historical balance table efficiently consolidates the day's movements. Leveraging UPDATE statements in BigQuery for thousands of hourly updates is inefficient and can lead to performance bottlenecks and increased costs due to full table scans or expensive row-level operations. Partitioning by item might help with some queries but doesn't address the core issue of frequent, small updates to a large table. Using the BigQuery bulk loader is suitable for large, infrequent loads, not for near real-time streaming updates.