Fujitsu released 
Databases rarely make headlines, and Fujitsu Enterprise Postgres isn't trying to. What it does is take proven, well-regarded open-source technology and do the hard work of making it genuinely enterprise-ready, with the enhancements, integration depth, and operational tooling that production environments actually demand. That combination of trusted foundations and substantive engineering is what defines this release.
 Fujitsu Enterprise Postgresunique feature |
 PostgreSQLextension |
|
 AI |
|
|
 Reliability |
|
|
 Security |
|
|
 Operability |
|
|
 Performance |
|
|
 DevOps |
|
|
Multi-Master Replication: Proven technology, enterprise depth
The ability to run multiple active database nodes, each accepting writes, with changes synchronized between them, is a well-understood architectural pattern. Fujitsu's implementation is built on pgactive, an open source extension developed by AWS, which itself is built on top of PostgreSQL's logical replication infrastructure. That's a foundation Fujitsu knows well, logical replication is an area where Fujitsu engineers have been active contributors to the PostgreSQL community for several years.
The result is an implementation that combines a mature, proven extension with a platform team that understands the underlying mechanics deeply.
But Fujitsu hasn't simply packaged pgactive and called it done.
Fujitsu Enterprise Postgres 18SP1 includes proprietary controls designed to deliver a more stable replication environment and integrate cleanly with other Fujitsu Enterprise Postgres capabilities, avoiding the rough edges that can appear when community extensions meet complex enterprise configurations.
The use cases this unlocks are meaningful:
- Multi-region active/active setups
In cases where you need genuine write availability in multiple locations rather than a warm standby waiting to be promoted. - Reduced DR infrastructure waste
All nodes are doing real work rather than sitting idle. - Faster recovery
Continuous synchronisation means catching up on missed changes — not waiting through a full resync.
Multi-master replication does introduce real considerations around conflict resolution, what happens when two nodes receive conflicting writes simultaneously, and this is worth factoring into your architectural planning. It's not a complexity-free option, but for organizations with the right requirements, the combination of proven replication mechanics and Fujitsu's enterprise controls makes it a serious production option.
Connection Manager: Smarter, more resilient
The enhancements to Connection Manager in Fujitsu Enterprise Postgres 18SP1 are a good example of Fujitsu adding genuine operational value on top of the underlying infrastructure.
Load balancing now goes beyond round-robin. When connections are unbalanced, the system automatically routes new connections to whichever standby has the lightest load, maximizing resource utilization without requiring application-level logic to manage it.
More importantly, Fujitsu Enterprise Postgres 18 SP1 handles the awkward moments in cluster topology changes. When a standby is added, removed, or promoted, existing read connections can become unbalanced or, more dangerously, pointed at a server that isn't yet fully synchronized. Version 18 SP1 addresses both problems; connections are redistributed at the start of the next query cycle, and standbys that haven't fully caught up are automatically excluded from read targets.
Applications don't need to build complex reconnection logic to handle these scenarios; the platform manages it.
Flexible infrastructure options
Fujitsu Enterprise Postgres 18 SP1 also extends support for open-source Patroni and etcd, enabling configurations with multiple replicas that deliver both automatic failover and improved read performance. And for cloud-hosted deployments, integration with Amazon EC2 Auto Scaling means replica counts can adjust automatically to match load, a practical operational improvement for teams running Fujitsu Enterprise Postgres in AWS environments.
AI Integration: Built on open standards, ready for production
Interest in connecting AI applications to enterprise data is high, and the pattern most organizations are exploring is Retrieval-Augmented Generation (RAG) where an LLM draws on your actual business data to generate relevant, accurate responses. Fujitsu Enterprise Postgres has supported the building blocks for this for some time, including vector storage and similarity search. Version 18 SP1 extends the platform in two directions that address real production concerns.
In-database embedding models
Generating the vector representations that make semantic search possible has typically meant calling an external model service OpenAI, Ollama, or similar. Fujitsu Enterprise Postgres 18 SP1 adds the ability to import ONNX-format embedding models directly into the database, running them locally. For teams with air-gapped environments, data sovereignty requirements, or a preference for fewer external dependencies in their architecture, this is a good option. It also simplifies the operational picture by keeping the embedding process within the database rather than coordinating across services.
MCP Server support
The Model Context Protocol is an emerging open standard backed by broad industry momentum for how AI applications securely and consistently access external data sources. Fujitsu Enterprise Postgres 18SP1 implements MCP server capability, meaning AI tools that support the protocol can connect to Enterprise Postgres without custom integration work for each one.
Crucially, Fujitsu has done more than implement the protocol. Fujitsu Enterprise Postgres 18SP1 ships with sample tool definitions for the patterns most likely to be useful such as semantic text search, hybrid search combining vector and traditional methods, and natural-language-to-SQL generation.
These aren't trivial additions, they give development teams a practical starting point that reflects real-world AI application patterns rather than leaving them to figure out the integration from scratch
The platform underneath it all
New features only matter if the platform they sit on is solid. Fujitsu Enterprise Postgres remains an enterprise-supported distribution of PostgreSQL, which means security patching, long-term support, and certified configurations that organizations can rely on across the product lifecycle. The Kubernetes operator continues to make containerized deployment and lifecycle management tractable for teams moving away from traditional database infrastructure.
For AI-adjacent workloads in particular, the data preparation side is frequently underestimated. The support by Fujitsu's Postgres for JSON, full-text search, and hybrid query patterns gives teams a flexible environment for shaping data into forms that AI workloads can use effectively without sacrificing the relational structure that transactional systems depend on.
Conclusion
Summing up, Fujitsu enterprise Postgres 18 SP1 is built on well-proven open source foundations, and the honest answer is that's a feature, not a limitation. PostgreSQL's ecosystem is deep, and leveraging the best of it, while adding the enterprise controls, operational tooling, and integration depth that production use actually requires is exactly what an enterprise distribution should do. Multi-master replication, smarter connection management, in-database AI model support, and MCP integration together make Fujitsu Enterprise Postgres 18 SP1 a release worth paying attention to, for the right reasons.
Take it for a spin and see for yourself
You can try Fujitsu Enterprise Postgres with a fully-featured trial version valid for 90 days to experience a frictionless hybrid cloud that can help you modernize to respond faster to business demands.
is an enhanced distribution of PostgreSQL, 100% compatible and with extended features.
PostgreSQL Insider has a series of technical articles for PostgreSQL enthusiasts of all stripes, with tips and how-to's.
