PGConf.Russia 2025

The issue of slow month-end closing and cost calculation on PostgreSQL has been a long-standing problem. In this talk, we will explore what modern versions of PostgreSQL offer in this regard and answer the question, "Do replicas slow down the master?"

We will also take a look at different types of physical replication and, for dessert, share a recipe for quick and painless month-end closing in 1C without affecting the daily operations of users.

The execution time of SQL queries depends on factors like indexes and up-to-date statistics. In most cases, optimizing slow queries helps resolve database performance issues.

But what if classic query optimization doesn’t work? What if the system keeps behaving unpredictably — or worse, crashes, leading to frustration, panic, and even despair?

In this talk, we’ll explore how Postgres Professional’s performance engineers tackle these challenges, look at the tools they use, what’s still missing, and where PostgreSQL performance optimization is headed

Although we often don't like to talk about it, database administration carries with it a large amount of stress. This is particularly true when we have to take on late night support duties. In this talk you will learn: 1. The basics of the physiology of stress. This is not a physiology lecture, but as engineers we need enough to build a mental model. 2. What kinds of mistakes we make when under stress. 3. How to measure stress in your own life and recognize it in others. 4. What tools are helpful to manage short- and long-term stressors. Managing short-term stress is important in avoiding becoming overwhelmed when problems occur, but managing longer-term stressors is just as important in avoiding problems in the first place.

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.