27:54

Félix GV, a former engineer at LinkedIn and architect of the Venice database system, discusses the complexity of building planetary-scale data systems. He explains Venice's unbundled architecture where each component—from Kafka-based pub/sub to RocksDB-powered servers—operates as an independent distributed system. Félix details their rigorous chaos engineering practices, including regular load tests that push data centers beyond normal capacity to ensure reliability. The discussion covers fundamental distributed systems concepts like the CAP theorem and the trade-offs between consistency and availability in multi-region deployments. He also explains why Venice, as a derived data system, deliberately sacrifices strong consistency for high throughput and availability, and concludes by discussing their experimental integration of DuckDB for SQL-based analytics and data exploration capabilities.