WILMINGTON, Del. — EnterpriseDB (EDB), the main sovereign AI and information firm, has introduced EDB Postgres Distributed (PGD) 6.4, the distributed transactional basis of EDB Postgres AI. The discharge introduces three landmark capabilities: Quorum Commit for true cross-node distributed consistency, built-in connection pooling by the native Connection Supervisor, and full assist for PostgreSQL giant objects. Collectively, they ship the sturdiness ensures and architectural simplicity that Tier 1 monetary and infrastructure purposes demand.
Demand for sovereign management over essential information is accelerating. Gartner forecasts that by 2030, greater than 75% of European and Center Japanese enterprises will geopatriate workloads — up from lower than 5% in 2025. AI and information sovereignty are clear strategic imperatives.
However management over the place information lives is simply half the image; organizations additionally want certainty about its state. PGD 6.4 meets that demand, bringing the strongest distributed consistency Postgres has ever supplied.
Quorum Commit: The tip of the distributed double-spend
Probably the most consequential addition in PGD 6.4 is Quorum Commit; a pre-commit coordination mechanism that enforces a single, coherent international fact throughout each node within the cluster earlier than any transaction is domestically dedicated. In contrast to conventional synchronous replication, which confirms solely {that a} reproduction has acquired a report, Quorum Commit coordinates the content material of concurrent writes throughout information facilities in actual time.
Contemplate a bank card shared throughout geographies. Beneath customary replication, two simultaneous fees can every go an area steadiness test and commit independently, leaving the financial institution to resolve an overspend after the actual fact. With Quorum Commit, the cluster reaches consensus earlier than both cost completes: The primary transaction wins the quorum vote and is dedicated; the second sees the up to date steadiness and is declined on the level of sale with none post-commit cleanup.
“Postgres has turn into the de facto customary for contemporary purposes. But till now, organizations working high-value workloads in banking, funds, or telecom had been compelled to fall again on legacy enterprise RDBMS for his or her strongest consistency necessities. PGD 6.4 adjustments that, delivering the identical distributed consistency these techniques had been constructed on, now totally Postgres native,” stated Jozef de Vries, SVP, Database Engineering at EDB.
For monetary establishments migrating away from proprietary databases, Quorum Commit removes the final main technical barrier: a totally Postgres-native path to the identical consistency ensures supplied by high-end industrial techniques with not one of the licensing overhead.
Connection Supervisor pooling: Enterprise scale, zero exterior dependencies
PGD 6.4 extends the built-in Connection Supervisor launched in PGD 6.0 with native connection pooling, eliminating the necessity for exterior poolers similar to pgBouncer in most manufacturing topologies. As a result of the Connection Supervisor is built-in immediately with PGD’s Raft consensus layer, it offers capabilities no exterior proxy can match: cluster-aware routing at each the cluster and area ranges, computerized failover routing that responds to consensus adjustments in actual time, and unified observability by PostgreSQL’s personal logging and monitoring views.
The architectural payoff is critical at scale. Organizations working lots of of distinct clusters similar to cost processors working throughout hundreds of CPU cores achieve an easier, extra resilient stack with one fewer element to deploy, model, and monitor. Operational complexity is decreased with none sacrifice in throughput or connection headroom.
Giant object assist: Broadening the scope of distributed Postgres
PGD 6.4 provides full replication assist for PostgreSQL giant objects—binary information constructions that had been beforehand unavailable in distributed Postgres environments. This brings a category of workloads into the PGD fold that had been beforehand excluded: purposes managing scanned paperwork, picture archives, binary payloads, or blended transactional and unstructured information inside a single database. Monetary establishments, authorities companies, and healthcare operators generally keep precisely these hybrid schemas alongside their core transactional tables.
PGD 6.4 is now out there as a stand-alone distribution and can be out there as a part of EDB Postgres AI in June. For technical documentation, launch notes, and improve steering, go to enterprisedb.com/docs/pgd.
