Google has announced Spanner Graph, an addition to Google Spanner that adds graph database capabilities to Google's fully managed distributed SQL database service.

Google describes Spanner as combining the best of relational and NoSQL databases - scalable, distributed, but supporting ACID transactions and SQL queries. Spanner is used internally at Google for many of Google's apps, including Gmail, Photos, and Calendar. It is available as a database as a service running on Google Cloud.

Google's internal implementation of Spanner is the largest single database on earth and spans all Google’s data centers - hence the name Spanner.

Spanner Graph provides a unified database that integrates graph, relational, search, and AI capabilities with virtually unlimited scalability.

Spanner Graph has an ISO Graph Query Language (GQL) interface that provides users with a concise way to match patterns and traverse relationships, based on open standards. Google says there is full interoperability between GQL and SQL to break down data silos and let developers choose the optimal tool for each query. They say that tight integration of graph and table data helps remove operational overhead and the need for complex and costly data movement.

Spanner Graph's built-in search capabilities combine vector and full-text search capabilities for graph data retrieval using semantic meaning and keywords. The combination means users can traverse relationships within graph structures using GQL, while simultaneously using search to find graph contents. Google says that specifically, developers can use vector search to discover nodes or edges based on semantic meaning, or use full-text search to pinpoint nodes or edges that contain specific keywords. From these starting points, they can then explore the rest of the graph using GQL.

Because Graph runs in Spanner, it benefits from Spanner's scalability, availability, and consistency. It also has deep integration with Vertex AI so developers can use the Vertex suite of AI models directly within Spanner Graph. Google says that for example, you can generate text embeddings for graph nodes and edges using LLMs, enrich your graph with the results, then use vector search to retrieve from the graph in the semantic space.

Google Spanner Graph is available now.

More Information

Spanner Homepage

Spanner Graph codelab

Spanner Emulator On GitHub

Related Articles

Google Spanner Wins Hall Of Fame Award

Google Releases Spanner Emulator

Google Spanner To Settle Relational Vs NoSQL Debate

Google's F1 - Scalable Alternative to MySQL