September 2021 Summaries

Vespa introduces a unique tensor formalism that differs from conventional machine learning frameworks by using named dimensions and supporting both sparse and dense tensor types, alongside a concise set of core functions. This approach enhances tensor computation by allowing semantic verification and static type inference, thus reducing runtime errors. Vespa's tensor framework employs a small set of foundational operations—such as map, reduce, and join—enabling complex computations without extensive libraries, which improves interoperability and maintainability. The framework supports dense, sparse, and mixed tensors, allowing flexible data representation and efficient space utilization. By leveraging named dimensions, Vespa provides a robust type system that simplifies tensor operations, making numerical computation more expressive and easier to optimize. The recently introduced tensor playground offers an interactive environment for exploring these tensors and tensor expressions.

Vespa's September 2021 product updates include several significant enhancements aimed at improving performance and usability. The Vespa CLI, a zero-dependency tool built with Go, simplifies interactions with Vespa instances across various operating systems, allowing users to clone, deploy, and query applications efficiently. Notable performance improvements include a 20x increase in exact nearest neighbor search performance without the HNSW index, reduced memory footprint for the HNSW index, and multithreaded reindexing capabilities. Additionally, Vespa now supports mTLS across internal services, enhancing security. Updates in feed performance have resulted in a substantial increase in throughput for both partial updates and summary-only data puts. The release also introduces the SentencePiece Embedder, allowing efficient transformation of natural language text into tensors for retrieval and ranking, with a flexible Embedder interface for custom algorithms. Vespa, largely developed by Yahoo engineers, continues to evolve through community feedback, serving as a robust open-source big data processing and serving engine used in various Yahoo products.

Vespa has introduced Vespa CLI, a zero-dependency command-line tool designed to simplify the deployment and interaction with Vespa applications, whether self-hosted or on Vespa Cloud. Historically, users interacted with Vespa applications using APIs or a Maven plugin, which required either terminal command expertise or a Java development toolchain. Vespa CLI, developed in Go and available on Linux, macOS, and Windows, eliminates these prerequisites by offering a more user-friendly alternative. It allows users to clone sample applications, deploy applications locally or remotely, interact with a dev zone in Vespa Cloud, feed and query documents, and send custom requests with automatic authentication. The tool is open-source, shares the same license as Vespa, and is available for installation via Homebrew or GitHub. Users are encouraged to explore the getting started guides and can provide feedback through GitHub issues.