Google is holding its Cloud Next conference in Tokyo this week, with a focus on tailoring its database for AI workloads (because at this point in 2024, AI is all these big tech companies want to talk about), which includes updates to its Spanner SQL database, adding support for graph and vector searches, as well as enhanced full-text search capabilities.
This wouldn't be a Google announcement without Gemini-powered features, which include Gemini for BigQuery and Looker to assist users with data engineering and analytics, governance and security tasks.
Google argues that while the majority of companies believe generative AI is critical to their business success, they also recognize that much of their data is ungoverned and outside the scope of their analytics and AI efforts.
“We need to move away from existing data silos and data islands and build an integrated multi-modal data platform that encompasses structured and unstructured data. [because] “GenAI excels at analyzing unstructured data, and it combines working with data at rest and data in motion — real-time data and data at rest,” explains Gerrit Kazmeier, vice president and general manager of databases, data analytics, and Looker at Google. Activating this enterprise data flow is what many of these new features aim to do, he argues.
Spanner Adds Graph and Vector Capabilities
Spanner powers most of Google's own products, including Search, Gmail, and YouTube, and its customers include Home Depot, Uber, Walmart, etc. While Spanner can handle vast amounts of data, vector and graph databases are essential for ingesting enterprise data into GenAI applications and enhancing existing underlying models.
“We're thinking about, what does it take to take Spanner's availability, scale, and relational model and actually extend it into the data platform of choice for running GenAI apps,” said Andi Gutmans, vice president and general manager of databases at Google. For Google, like many database vendors, the first step is to add graph capabilities to Spanner using the emerging GraphQL standard. Companies can then use this graph to extend their GenAI applications and the underlying models that power them with Retrieval Augmented Generation (RAG), which is now the de facto standard for this use case.
Spanner's new features also include full-text search and vector search, with the vector search feature supported by Google's ScaNN algorithm. “With Spanner Graph, full-text search, and vector search, Spanner is not only the most highly available, globally consistent, and scalable database, but also a multi-model database with intelligent capabilities that can seamlessly interoperate to power a new class of AI-enabled applications,” Google said.
In addition to these AI-centric updates, Spanner is also introducing new optional pricing. Called “Spanner Editions,” the concept is to offer a more flexible, tier-based pricing model. Today, Google Cloud customers have to choose between a single-region edition and a multi-region edition that also bundles additional features such as replication.
Bigtable now supports SQL
Google on Thursday also announced a major update to Bigtable, Google's NoSQL database for unstructured data and latency-sensitive workloads: Bigtable now features SQL support (or, more precisely, support for Google's own SQL dialect, GoogleSQL), making it much easier for virtually any developer to use the service.
Previously, developers had to use the Bigtable API to query the database. Currently, Bigtable supports about 100 SQL functions.
Oracle on Google Cloud
For Oracle database fans, Google will allow Oracle Exadata and Autonomous database services to be hosted in Google Cloud data centers, and applications can be linked between Google Cloud and Oracle Cloud. For Google, this means more workloads in the cloud, and for Oracle, at least, these users will continue to pay license fees even if they are not using Oracle Cloud.
Also new for Google Cloud is support for open source Apache Spark and Kafka for data streaming and processing, as well as real-time streaming from Analytics Hub, Google's service for securely sharing data between organizations.