Google Cloud beefs up database services for generative AI workloads – SiliconANGLE

Google Cloud said today it’s updating multiple database services, including AlloyDB, Spanner and Memorystore for Valkey 7.2, with new features that should make them more suitable for generative artificial intelligence applications.

To begin with, Google announced that it’s making the ScaNN vector index generally available in AlloyDB, which is the company’s fully managed PostgreSQL-compatible database service, aimed at high-performance database workloads.

In a blog post, Andi Gutmans, Google Cloud’s general manager and vice president of engineering of databases, explained that the ScaNN vector index is the same technology that powers services such as Google Search and YouTube. The ScaNN technology is designed to make unstructured data such as videos searchable by representing it as vector embeddings.

The integration means that AlloyDB now supports the same, powerful vector search capabilities used by Google’s own search engine, making it much more useful as a foundation for generative AI applications.

In addition, the company is building on last year’s launch of AlloyDB Omni, which is a downloadable version of AlloyDB that can run on any platform and in any location, including on-premises or in other clouds.

Today, it said it’s partnering with a company called Aiven Ltd. to launch a fully managed version of AlloyDB Omni. With Aiven for AlloyDB Omni, customers can use a managed cloud database that can easily be deployed and managed on any cloud platform, including Amazon Web Services and Microsoft Azure, making the database much more useful for customers with multicloud environments.

“You can now run transactional, analytical and vector workloads across clouds on a single platform, and easily get started building gen AI applications, also on any cloud,” Gutmans said.

Vector search comes to Memorystore databases

In addition to AlloyDB, Google is also enhancing its managed services for Valkey, an open-source fork of the popular Redis database – an in-memory database that also functions as a data cache, streaming engine and message broker.

Just last month, Google debuted a new service called Memorystore for Valkey 7.2, essentially providing a fully managed experience for the Valkey 7.2 database. It sits alongside Google’s Memorystore for Redis Cluster managed services, and today, both offerings are being updated with support for rapid vector search capabilities. According to Gutmans, a single Memorystore for Valkey or Memorystore for Redis Cluster instance can now perform vector search with single-digit millisecond latency on more than a billion vectors, with more than 99% recall.

The vector search capabilities are integrated with Memorystore for Valkey 7.2, and can also be found in Memorystore for Valkey 8.0, which is a managed version of the latest Valkey release, available in public preview from today. According to Gutmans, Memorystore for Valkey 8.0 brings some serious performance improvements, achieving twice as many queries per second compared to Memorystore for Redis Cluster.

Firebase gets its first relational database

Moving on, Google unveiled an entirely new database service called Firebase Data Connect, which is said to be the first relational database service designed to work with the Firebase application development platform.

Gutmans explained that Firebase Data Connect is a backed-as-a-service integrated with a fully-managed PostgreSQL database that’s powered by CloudSQL. It’ll launch in public preview later this month, and will support features including rich queries, complex conditions and semantic vector search to power advanced generative AI applications for Android, iOS, web and Flutter.

Fleet management enabled for Google Spanner

As for Google Cloud Spanner, it didn’t receive any major new features today, but users may still be excited to learn that it’s being added as a supported database in the company’s Database Center, alongside the Cloud SQL and AlloyDB databases.

Cloud Spanner is a kind of hybrid database that’s able to store relational data organized in rows and columns, as well as semistructured data such as business documents and product description collections. In addition to its versatility, it’s known to be among the most reliable databases around, with various features that ensure its high availability.

Meanwhile, the Database Center is a relatively new service launched by Google earlier this year that enables information technology administrators to manage entire fleets of databases side-by-side. To do this, it leverages the skills of Google’s Gemini generative AI models to help monitor tens, hundreds or even thousands of individual database deployments, reducing risks and improving their performance. It monitors Google’s various database offerings through a single interface that spans various cloud regions, making life much simpler for admins.

Expanded availability for Oracle Database@Google Cloud

Finally, Google said its Oracle Database@Google Cloud service is becoming more widely available with the official upcoming launch of a new South America East (Sao Paulo) region, which will join its existing regions in US East (Ashburn), US West (Salt Lake City), UK South (London), and Germany Central (Frankfurt).

The new region in Sao Paulo will be the first of many, as Gutmans revealed that the company has plans to more than double the number of regions that support Oracle Database@Google Cloud by the end of next year.

Oracle Database@Google Cloud is a modernized version of Oracle Corp.’s iconic database, customized to run on Google’s cloud infrastructure and enhanced with various features that make it more suitable for generative AI applications.

Image: SiliconANGLE/Microsoft Designer