Published 3 Mar 2026
Duration: 1718
The architecture of the Venice database system, used by LinkedIn, balances scalability and consistency through an asynchronous ingestion model and eventual consistency, prioritizing availability and partition tolerance.
This interview was recorded for GOTO Unscripted. https://gotopia.tech Check out more here: https://gotopia.tech/articles/421 Felix GV - Current Intere...
The podcast details the architecture of the Venice database system, a modular, distributed solution designed for recommendation data storage at LinkedIn. It emphasizes the system's unbundled structure, with distinct components such as a write-ahead log, a server fleet utilizing RuxDB, and specialized clients for querying, caching, and real-time data streaming. This design prioritizes scalability and flexibility, enabling efficient handling of large-scale data workloads. The discussion contrasts Venices asynchronous ingestion and eventual consistency model with traditional databases like PostgreSQL, which emphasize immediate consistency, highlighting the trade-offs between scalability and data consistency in distributed systems.
The systems resilience is evaluated through load simulations and chaos engineering techniques, such as those inspired by Netflixs Chaos Monkey, to test reliability during data center outages. The podcast also addresses the implications of the CAP theorem, noting Venices focus on availability and partition tolerance over strong consistency, particularly in multi-region architectures where reliability often takes precedence over strict consistency guarantees. Additional topics include data ingestion strategies, the use of derived data systems for optimization, and experimental integrations like DOXDB to enhance querying capabilities, underscoring ongoing efforts to adapt the system for evolving use cases.
19 Jun 2026 Continuous Delivery in a World of Constant Change Abby Bangser & Dave Farley
Continuous delivery principles, AI's challenges in code generation, and the necessity of incrementalism, rigorous testing, and human validation in software development are emphasized, alongside critiques of AI's lack of precision and the push for structured validation in high-stakes systems.
16 Jun 2026 Go for Java Programmers Barry Feigenbaum & Shon Saliga
Go emphasizes simplicity, concurrency, and efficiency for lightweight applications through explicit error handling and goroutines, while Java offers a feature-rich, object-oriented framework with extensive libraries and inheritance for complex, general-purpose systems, each suited to distinct development priorities and domains.
12 Jun 2026 Engineering Leadership in Turbulent Times Sarah Wells, Pat Kua & Daniel Terhorst-North
The text contrasts technical leadership and management, highlighting leadership's emphasis on vision and team alignment with management's focus on systems and accountability, while exploring change strategies, frameworks like the "Three Threes Model," challenges in technical debt and cross-functional alignment, and the role of communication, culture, and adaptability in fostering innovation.
9 Jun 2026 Modern Concurrency in Java Bazlur Rahman & Michael Redlich
Modern Java concurrency explores the shift from traditional threads to virtual threads (JDK 21) and structured concurrency, emphasizing scalability for I/O-bound tasks, task management simplification, limitations in CPU-bound work, comparisons with reactive programming, adoption challenges, and the book's unification of knowledge around Project Loom and scoped values for intermediate developers.
5 Jun 2026 Roc & Zig: A Compiler Rewrite Story Anjana Vakil & Richard Feldman
The text covers Rock's evolution as a simplified, statically typed alternative to Elm with a Zig-based compiler, AI's expanding role in software development beyond automation, open-source challenges, education's shift toward conceptual understanding, and the tension between rapid AI-driven productivity and quality-focused project development.