February 2025 Summaries

Vector search is a modern technique for finding the most similar items in a dataset by evaluating multiple dimensions or preferences simultaneously, rather than relying on traditional database queries that filter data based on specific conditions. This approach calculates the "distance" between items in a multi-dimensional space to determine the closest match, making it effective for scenarios like finding a compatible roommate at a conference or discovering a celebrity with similar traits. SurrealDB facilitates vector search by representing each set of preferences as a vector, enabling users to find matches by calculating distances using methods such as cosine similarity. This method is adaptable to various contexts and can handle numerous dimensions, revealing the closest matches through a structured mathematical framework.

The blog post explores the automation of knowledge graph construction using SurrealDB and the Gemini large language model (LLM), highlighting how these tools facilitate the representation and querying of complex, interconnected data. Knowledge graphs, which create networks of understanding by linking data relationships, have traditionally been labor-intensive to build; however, LLMs like Gemini can automate the extraction of entities and relationships from text, simplifying this process. SurrealDB, a multi-model database, supports both graph structures and vector embeddings natively, enabling sophisticated analysis and semantic search by capturing the meaning and relationships within data. The post demonstrates how to generate SurrealDB queries using Gemini to build a knowledge graph from an AI-generated text and suggests adding embeddings for enhanced semantic search capabilities. The integration of Retrieval Augmented Generation (RAG) techniques with knowledge graphs is presented as a powerful application, combining the data model and querying strengths of SurrealDB to improve the accessibility and utility of knowledge-driven applications.

In 2025, SurrealDB is focusing on enhancing performance, stability, and features, following a year of foundational development with releases such as Surrealist and SurrealDB 2.0. The latest release includes improved benchmarking to evaluate the database's performance against other databases, significant upgrades in testing strategies for their query language SurrealQL, and the addition of graph path algorithms for better data handling. Notably, experimental features like record references introduce referential integrity similar to SQL's foreign key constraints, allowing more sophisticated data relationships. Other enhancements include better HTTP error messages, new environment variables for monitoring, optimized WebSocket implementation, and a new function to check object emptiness. The updates are aimed at improving developer experience and system efficiency, with comprehensive release notes available for further details and community feedback encouraged.

SurrealDB is a versatile multi-model database that integrates various data types, such as relational, document, graph, and time-series, while supporting diverse deployment environments from embedded devices to distributed clusters. By separating the storage and computation layers, SurrealDB blurs the lines between traditional databases and backends as a service, incorporating built-in security and API layers for REST, RPC, and GraphQL. To benchmark its performance across different configurations and databases, the developers have created an open-source tool called crud-bench, which is designed in Rust and can be customized for various workloads, including SQL and NoSQL platforms. The tool focuses on testing features unique to SurrealDB, allowing developers to assess the impact of changes on database performance. SurrealDB's benchmarking efforts, which include regular tests and a call for community contributions, aim to optimize performance and functionality while striving for fairness in configuration comparisons with other databases.

The new release of the Python SDK for SurrealDB introduces several enhancements aimed at improving performance, consistency, and integration capabilities. This update ensures the interface for WebSockets and HTTP requests aligns with other SurrealDB SDKs, such as the JavaScript SDK, and supports both blocking and asynchronous implementations to cater to a variety of user needs. Notably, the SDK transitions from JSON to CBOR serialization, offering performance advantages and flexibility with unstructured data, while also supporting native Python datetime objects for seamless integration. The release emphasizes stability with 100% test coverage across isolated classes for each request method, resulting in 5,675 tests covering Python versions 3.9 to 3.13 against various SurrealDB versions. Additionally, the SDK facilitates easy contribution by allowing users to download the source code and run tests in a debugger to observe data transformations in real-time, promoting transparency and ease of use.

SurrealDB has been employed to address the challenges of patient recruitment in clinical trials by creating a more efficient and connected drug trial search engine. Clinical trials, crucial for assessing new treatments, face significant recruitment hurdles due to high costs and patient unawareness. SurrealDB's multi-model architecture, integrating graph, document, table, and vector-based storage, allows for efficient modeling and querying of complex relationships among drugs, disorders, and trial data without the need for multiple databases. This approach reduces complexity, enhances search capabilities, and improves scalability, making it easier for researchers and patients to access relevant trials. The system's unified query language, SurrealQL, simplifies data interaction, leading to faster queries and streamlined development processes, ultimately aiming to lower recruitment costs and expedite the availability of new treatments.