Google Cloud makes Spanner Omni generally available: its Spanner database can now run in companies' own data centers and on other clouds

What happened

Google Cloud announced on its official blog that Spanner Omni, the deploy-anywhere version of its Spanner database, is now generally available (GA). In plain terms, it has left its preview stage and is offered for production workloads, the live systems that serve real customers, in a company's own data center or on other clouds. Google Cloud says Spanner Omni debuted at Google Cloud Next ’26.

Spanner is a distributed database: it spreads data across many servers. According to Google Cloud, Google pioneered the distributed SQL market with Spanner more than a decade ago, combining the horizontal scalability of NoSQL (growing by adding more machines) with the ACID compliance and strong consistency of traditional relational databases (guarantees that changes are saved reliably and everyone sees the same up-to-date data). Google Cloud also says Spanner has evolved into a multi-model database that combines SQL, graph, key-value, full-text search, vector search and analytical processing using a columnar engine in a single database.

Read the full description

Sources are collected, independently checked, then reviewed by Jev.

Until now, Spanner was offered as a fully managed service, meaning Google runs it for customers on Google Cloud. Google Cloud says Spanner Omni offers the same core capabilities and can be deployed on virtual machines or Kubernetes (software for running applications in containers) in private data centers, across multiple clouds, or run locally on a laptop for testing.

Key features announced by Google Cloud

• AI capabilities: Google Cloud says it supports vector search, which finds items by similarity of meaning (including exact KNN and approximate ANN nearest-neighbour search), and Spanner Graph. It also supports the Model Context Protocol (MCP), an open standard for connecting AI agents to data, via MCP Toolbox, letting AI agents inspect data schemas and retrieve relevant context.
• Security and governance: According to Google Cloud, the GA release provides TLS encryption, authentication and authorization, and audit logging.
• Data protection: Google Cloud says backup and restore are available.
• Worker nodes: Google Cloud describes these as stateless compute nodes (they do not store the database's data) specific to Spanner Omni, used to offload background and resource-intensive operations.
• Support: Google Cloud says customers can use Google Cloud Customer Care directly.

Google Cloud also cites a customer example: Attio, a London-based AI-native CRM (customer relationship management software) company, has migrated its agentic application environment to containerized Spanner Omni plus managed Spanner. The blog quotes Attio co-founder and CTO Alexander Christie: “Spanner Omni has been a major win for us by delivering Spanner capabilities and performance across all our environments”.

Licensing and differences from the fully managed version

Google Cloud says Spanner Omni comes in two licences. The free Developer Edition is for non-commercial, non-production use; its default licence lasts 90 days and includes all Commercial Edition features except backups and Worker nodes. For single-server deployments with 4 vCPUs (virtual processor cores) or fewer, the default licence does not expire and supports backup and restore; beyond 4 vCPUs, an extension can be requested via a form. The paid Commercial Edition uses an annual subscription priced per vCPU and includes enterprise support, and a discounted proof-of-concept licence is also available.

What it means for everyday readers and businesses

For ordinary users, databases like this mostly run behind the scenes and won't directly change the apps they use every day. But it reflects a trend: cloud providers are starting to let enterprises run core services, once offered only on their own clouds, in their own data centers or on other clouds, so organizations constrained by regulations, data residency or existing infrastructure can adopt them too.

On the other hand, according to Google Cloud, running it yourself shifts operational responsibility to the customer, with no availability SLA, the uptime guarantee, from Google. Businesses evaluating it need to weigh their in-house team's ability to run it and whether they need integrations such as BigQuery that are only available in the fully managed version.

Browse the latest news in this topic

• Tech news