PGConf.Russia 2025

In PostgreSQL performance testing, benchmarks measure query execution time (latency). To get more reliable results, queries are executed repeatedly, generating a dataset of latency values. Performance is often assessed using standard metrics like the median or mean, but we propose a more advanced approach.

In practice, latency distributions are often multimodal, consisting of multiple underlying distributions with distinct characteristics. In such cases, traditional statistical methods are insufficient, requiring a more detailed analysis of the dataset’s structure.

Our work presents a tool that automatically performs statistical analysis of benchmark results, accounting for dataset-specific features. It detects multimodality, identifies the number and boundaries of dominant modes, and determines key distribution parameters—providing deeper insights into PostgreSQL performance variations.

The visibility map in PostgreSQL is an important mechanism for optimizing database performance, accelerating read operations, data cleanup, and indexing. Despite some limitations, such as memory consumption and fragmentation, its use can significantly enhance performance in high-load systems. Effective utilization of the visibility map requires proper configuration and monitoring, which is especially crucial in systems with large data volumes and high transaction concurrency.

In this talk, I will challenge architectural stereotypes about the impossibility of communication with external systems directly from a DBMS. PostgreSQL offers vast communication capabilities; you just need to know which extensions to use.

And when extensions aren't enough, you can create your own. For the Luxms BI project, we needed communication with NATS, so we developed our own extension in Rust. I’ll share the GitHub link and my experience of developing PostgreSQL extensions in Rust during this talk.

For over a year, Postgres Pro has provided extended vacuum statistics, reflecting its operation on individual relations.

We've started receiving observations from production databases of clients that include these statistics. It has been quite successful, and in 2024, we began actively promoting vacuum statistics in PostgreSQL. In this talk, we will review what these statistics can tell us about the complex life of vacuum, using real-world data from live systems.