новости компании о DBtune autonomous optimization blows PostgreSQL bottlenecks away

Startup DBtune builds autonomous PostgreSQL database‑optimization capabilities with potential to lift PostgreSQL performance to compete against Oracle.

Headquartered in Malmö/Lund, Sweden, DBtune was founded in 2020 by machine‑learning researcher Dr. Luigi Nardi as a spin‑out from Stanford University and Lund University. The firm secured startup grants from the Wallenberg Foundation and Vinnova across 2020‑2021, and closed a $2.4 million seed funding round in July 2023 backed by venture capital firms and angel investors.

Its solution leverages agentic AI to tune databases for individual workloads, use‑cases and underlying hardware via three core functions: server‑parameter adjustment, index refinement and autovacuum oversight. DBtune evaluates PostgreSQL server‑parameter sets against real‑world workload and machine profiles, applying optimal configurations automatically or awaiting user approval. DBtune v4.0 employs ML‑driven AI to dynamically adjust server settings, suggest suitable indexing schemes and track autovacuum activity aligned with given workloads and hardware limits.

Dr. Luigi Nardi commented: “As PostgreSQL deployments scale, preserving performance and stability demands more expert labor, manual adjustments and infrastructure resources. Meanwhile enterprises face mounting cloud‑cost pressure while modernizing stacks for growing AI adoption. We see Autonomous PostgreSQL as the next evolution, where agentic AI cuts down repetitive operational database tasks.”

He noted: “Realizing this needs a re‑imagined database‑optimization methodology. DBtune has invested ten‑year research plus multiple product‑development cycles toward this goal. DBtune 4.0 moves past pure server‑parameter tuning, unifying index optimization and autovacuum monitoring under its agentic‑AI‑powered autonomous PostgreSQL workflow.”

The platform detects and prioritizes index opportunities according to workload patterns, displaying projected impact, cost and corresponding DDL statements. Output covers proposed new indexes for accelerating critical queries, plus existing indexes requiring maintenance from bloat or removal for obsolescence and duplication. DBtune also spots autovacuum backlogs and assists root‑cause diagnostics.

DBtune customer Midwest Tape is a full‑service media distributor supplying physical and digital content solely to public libraries throughout North America, Australia and New Zealand. The company runs a dedicated PostgreSQL replica to feed metadata for its Hoopla digital‑content offering to library clients. Deploying DBtune for AWS RDS PostgreSQL tuning resolved query bottlenecks and delivered a 10.8‑fold performance uplift; average query latency dropped from 75.9 ms down to 7 ms.

DBtune completed benchmark testing using Intel Xeon hardware to quantify performance gains from autonomous tuning versus thorough manual parameter work, yielding three key outcomes:
• Transaction throughput rose 87 percent within the initial optimization cycle
• An additional 17‑percent throughput gain occurred on top of an already heavily manually‑tuned environment
• Disk I/O fell 43 percent alongside a 23‑percent reduction in WAL volume generation

Without modifying application logic, schema layout or physical hardware, DBtune autonomously pinpointed major bottlenecks, verified root causes via iterative testing and translated insights into tangible performance and efficiency improvements. That 17‑percent throughput gain on pre‑tuned infrastructure proves autonomous tuning can unlock latent performance even after conventional tuning addresses obvious low‑hanging fruit.

Complete autonomous PostgreSQL remains a work‑in‑progress. Even so, Nardi contends surging cloud expenses and rising complexity from expanding applications and datasets make AI‑led database tuning increasingly vital. While PostgreSQL holds solid enterprise traction, substantial source‑code modifications will be necessary for native autonomous capabilities. Should the open‑source community execute these changes well, PostgreSQL could establish competitive advantages over proprietary alternatives such as Oracle.

Beijing Qianxing Jietong Technology Co., Ltd.
Sandy Yang/Global Strategy Director
WhatsApp / WeChat: +86 13426366826
Email: yangyd@qianxingdata.com
Website: www.qianxingdata.com/www.storagesserver.com
Business Focus:
ICT Product Distribution/System Integration & Services/Infrastructure Solutions
With 20+ years of IT distribution experience, we partner with leading global brands to deliver reliable products and professional services.
“Using Technology to Build an Intelligent World”Your Trusted ICT Product Service Provider!