The cloud-native storage engine for PostgreSQL
OrioleDB is a PostgreSQL extension that combines the advantages of both on-disk and in-memory engines. It uses PostgreSQL pluggable storage to increase performance and cut costs.
Scalability
The novel design of OrioleDB avoids typical bottlenecks of database engines.
IO Reduction
Logical WAL and non-persistent undo log for less IO from write transactions.
Data Compression
Page-level data compression reduces typical database size in 4-5 times.
Bloat Reduction
Less bloat thanks to efficiently handled in-place update.
Less Maintenance
Lower maintenance required resulting in less operational costs.
High Transaction Throughput
Efficient transactional processing leads to high throughput.
Scalability
OrioleDB utilizes the power of modern hardware to achieve extraordinary performance.
On a read-only nine-key point lookup OrioleDB-beta18 reaches over 1.25M TPS at 100 clients — more than 5× the throughput of stock PostgreSQL 18.6 on the same c5d.metal hardware.
On the classic pgbench read-write transaction wrapped in a plpgsql function OrioleDB-beta18 sustains over 270k TPS, roughly 2.7× the plateau of stock PostgreSQL 18.6 and with a much wider stable region.
Efficiency
OrioleDB's architecture dramatically reduces resources used for transaction processing.
OrioleDB keeps the same hour of vehicle-tracking data in 17 GB of tables and indexes, where PostgreSQL 18.6 needs 30 GB. The saving comes from index-organized tables: rows live in the primary-key tree itself, so no separate index duplicates them.
Under the same workload OrioleDB issues 2.9× fewer disk writes than PostgreSQL 18.6 — 99 against 284 IOPS on average.
OrioleDB is a PostgreSQL extension that combines the advantages of both on-disk and in-memory engines. It uses PostgreSQL pluggable storage to increase performance and cut costs.