Extended Amazon Aurora, new local demo benchmark and more
We’ve been busy fine-tuning our platform to give you more control, flexibility, and a smoother onboarding experience. Here is a look at the latest features we rolled out over the past month.
Extended Amazon Aurora optimization: We have expanded our parameter optimization space for Aurora to better tune complex join operations: apg_adaptive_join_cost_threshold and apg_adaptive_join_crossover_multiplier, which are available since PostgreSQL 17.4. Learn more about the new optimization space.
New local demo benchmark: Want to test-drive DBtune before deploying it in your enterprise environment? Now, when you create a DBtune account, you have access to a built-in demo database benchmark. This allows new users to safely and easily experience the DBtune platform directly from their desktop or laptop machines on Linux, macOS, and Windows (via WSL).
Tuning session scheduler: Gain more control over your optimization workflows. With our new scheduling tool, you can now set up tuning sessions to run automatically at a specific date and time in the future.
DBtune Fellow Marc Linster's analysis of over 250 PostgreSQL tuning sessions from Q1 2026 reveals a significant "performance gap" in managed cloud services. Despite being "pre-optimized" with workload heuristics, services like Amazon RDS and Aurora saw gains between 1.7x and 2.7x in average when agentic AI replaced generic provider and user defaults with workload-specific parameters.
The first in our series "Autonomous PostgreSQL", Marc Linster sits down with DBtune CEO Luigi Nardi to discuss the evolution of AI-driven database optimization. This agentic AI approach allows for multi-dimensional tuning of interconnected variables, achieving up to a 10x performance boost and 50% lower CPU usage. By automating tedious tasks like parameter tuning and indexing, DBtune enables developers to move toward a truly self-driving database experience.
We welcome Profelis, Proventa, and Kangaroot to our global partner network. These industry leaders bring deep expertise in open-source technologies, digital transformation, and mission-critical infrastructure across Turkey, Germany and Belgium.
Find out about how our partners delivers PostgreSQL performance optimization solutions worldwide.
Join our Founder and CEO, Luigi Nardi, at the next AZUG meetup in Copenhagen for a hands-on deep dive into Azure Database for PostgreSQL flexible server optimization. In this practical workshop, you’ll learn which parameters matter most and gain experience using open-source tools like PGtune to drive performance. The session concludes with an introduction to DBtune Community Edition, our AI-powered engine that blends heuristic rules with ML for smarter, workload-aware optimization spanning server parameters and indexes. RSVP to secure your spot!
DBtune is Gold Sponsor at PostgreSQL Conference Europe 2026
Join us in Valencia, Spain this coming fall and stay tuned (no pun intended!) for exciting updates. We can't wait to connect with the global PostgreSQL community at the biggest PostgreSQL conference yet! Read more about the conference.
Upcoming talks and webinars
PGDay France (4 June): Workload Fingerprints: The precision metric for autonomous PostgreSQL tuning (Luigi Nardi)
POSETTE, online (17 June): Building safety tooling for risk-free AI tuning of Postgres: Fast cars need fast brakes (Mohsin Ejaz)
New hire
We welcome Justin Bauer as our new UX Designer! Justin comes from the B2B SaaS world, most recently at Klimato, where he led their design initiatives, built their design system from the ground up, and redesigned how teams work with sustainability data. Before that, at Appspotr Studios, he focused on accessibility and no-code tooling for mobile products across a range of industries. What stood out to us about Justin is his ability to translate complex technical workflows into interfaces that actually make sense to users, which is exactly the challenge we're solving at DBtune. We're excited to see what he builds with us.
Articles we love
The first draft of the PostgreSQL 19 release notes is available by The PostgreSQL Global Development Group PostgreSQL 19 introduces SQL/PGQ for property graph queries and ON CONFLICT DO SELECT for atomic get-or-create operations. Key operational upgrades include REPACK for online table rebuilding and parallel autovacuum. This release also removes the 32-bit MultiXact ceiling, eliminating a notorious source of production outages for high-concurrency workloads.
by Bruce Momjian This presentation by Bruce Momjian explores using PostgreSQL to support Model Context Protocol (MCP) workloads. MCP provides a standardized way for language models to securely retrieve external data and perform actions, overcoming traditional LLM limitations like static training data and imprecise outputs.
What's the worst PostgreSQL configuration mistake you've seen in production? by Philip Clarence He highlights that PostgreSQL defaults are intentionally conservative, often leading to significant performance "horrors" in production. By failing to tune key settings like shared_buffers and work_mem, or disabling autovacuum, systems suffer from massive disk spills and table bloat.
Why developers are betting on Postgres for AI by Meredith Shubel Meredith explores how PostgreSQL serves as a reliable "ground truth" for enterprise AI. By combining pgvector for similarity search with structured data in a single database, organizations can ground AI responses in accurate, live operational reality.
Stay in touch
Your direct line to ask questions, share insights, connect with the team and other DBtune users on Slack.
.png?upscale=true&width=1120&upscale=true&name=11th%20newsletter%20(1).png)
.png?upscale=true&width=1120&upscale=true&name=3%20months%20of%20tuning%20(2).png)
.png?upscale=true&width=1120&upscale=true&name=Slide%2016_9%20(1).png)
-1.jpg?upscale=true&width=1110&upscale=true&name=event-flyer%20(1)-1.jpg)
-1.png?upscale=true&width=1110&upscale=true&name=PGConf%20Europe%20(2)-1.png)
