Vespa Newsletter, November 2021

Vespa's November 2021 newsletter highlights several significant updates and improvements to its open-source big data processing and serving engine. Schema inheritance is now enhanced, allowing schemas to inherit all content from a parent schema, not just document types, and structs can now also inherit from each other. Data export performance has been improved with features such as streaming HTTP responses and slicing to enable parallel data access, while query performance has been optimized with new configuration options. Deployment of large Vespa application packages, particularly those using large machine learning models, is now faster due to optimizations that include parallel evaluation of rank profiles. Additionally, the support for bitwise Hamming distance has been added in ranking expressions and nearest neighbor searches, and the company has shared insights into the neural search paradigm shift through a presentation at Glasgow University. Vespa, largely developed by Yahoo engineers, continues to be an essential tool for various Yahoo platforms, with ongoing growth driven by community contributions and feedback.