If you’re wondering how Postgres is evolving, and what the future holds, you won’t want to miss our new Builders podcast series. In episode one, Peter Eisentraut, EDB Chief Engineer VP and Core Postgres Contributor, highlights the newest innovations in Postgres 16. He also gives us a glimpse into the future of Postgres and how we can leverage the power of AI.
Listen to the full podcast here.
Here are some of the highlights:
Eisentraut kicks off our session by noting how there’s been two main shifts over the many years he’s been involved with Postgres – one on the development side and one on the usage side. When it comes to development, Eisentraut says that Postgres development has professionalized, with more full-time developers coding on Postgres than ever before. On the usage side, more people are using databases in the cloud, and cloud providers are adopting and tweaking the software in their environments. These changes can be challenges but also opportunities for developers of open source, and Eisentraut explains why in the podcast.
What’s happening with Postgres today?
Postgres itself is moving forward steadily, with Postgres 16 featuring improvements in performance, usability, monitoring and JSON support. Eisentraut says it’s interesting to observe how every release is worth upgrading to, when you take the continuous improvement across different areas into account.
For future versions of Postgres, work is progressing on temporal tables, graph databases, a GQL database language and ways to combine relational with graph databases. Postgres’ extensive developer community continues to contribute to Postgres releases through systems like Commitfest, where patches are reviewed, improved, and committed.
Eisentraut covers some of the newer Postgres database extensions in the podcast, like EDB’s PG Failover Slots. Not all developers know about this open source extension that helps manage logical application failover. But everybody should consider using PG Failover Slots, notes Eisentraut, especially if you have complex setups involving physical and logical replication.
Another valuable extension is PostGIS, which allows you to store geographic information systems (GIS) objects in an EDB Postgres Advanced Server database. It includes functions for analyzing and processing GIS objects and support for GiST-based R-Tree spatial indexes. Pgvector is yet another important extension, which allows you to store, query and index vectors. While it’s not new, Pgvector has become super popular recently.
Data security is critical for all businesses, and Eisentraut discusses security features that help EDB customers ensure their data is well managed, protected, and tracked. Two of our main projects are Transparent Data Encryption (TDE), which encrypts storage on disk, often called encryption at rest; and column encryption, where the encryption happens transparently on the client so the server doesn't see the data in plaintext at all.
How will Postgres leverage AI in the future?
Eisentraut says the time has come for automatic, machine learning-driven management of Postgres. Machine learning could be used to configure memory, set up shared buffers, estimate costs and perform other essential database configuration and management tasks. While we’ll still need DBAs, the DBAs won’t have to decide whether shared buffers should be four gigabytes or five gigabytes – the machine can determine that.
Machine learning can also classify SQL queries and protect against SQL injection attacks. DBAs constantly optimize queries and catch the wrong ones, but machine learning can help with this as well as assessing and monitoring IO, memory usage and other statistics. While machine learning can drive meaningful improvements, our industry will still need to keep DBAs in the loop to make some of the more important decisions.
To hear more about where Postgres is going and get tips to drive your database projects and career, listen to the complete podcast here.