PGConf.Russia 2025

This talk explores the challenges we encountered — and solved — while migrating massive databases from Oracle to PostgreSQL.

One of the most complex aspects of this process was rebuilding foreign keys (FKs). To overcome these challenges, we had to dive deep into the internal workings of FK creation and validation.

Key topics include:

Does ALTER TABLE ... ADD CONSTRAINT ... FOREIGN KEY have an execution plan?

Can an ordinary user influence this process?

What locks are applied during FK creation and validation?

Also, we’ll introduce a first-principles method for investigating performance issues, applying it to a real-world FK creation bottleneck.

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.

Covering all the key changes in PostgreSQL 18 during this talk would be quite a challenge — especially since the code freeze is set to happen just a week after the presentation. However, I’ll focus on what has already been finalized.

There are plenty of exciting updates in performance, monitoring, vacuuming (of course!), and beyond. As always, this talk is based on a series of articles reviewing the commit fests for PostgreSQL 18, published on Postgres Professional’s corporate blog on Habr.

The focus of the presentation is on the challenges of building an application solution on PostgreSQL, specifically a high-load analytical data warehouse. Using the case of the company EVRAZ, we will demonstrate the impact of applying the Data Vault methodology on PostgreSQL and Greenplum databases for developing a unified production performance system.

We will discuss the difficulties and solutions, showcasing query plans for tasks such as updating directories and handling large objects, along with optimization examples. We will highlight coding nuances, problems related to populating the model, and issues with querying from the model.

This session will be useful for those planning to use or already facing challenges with the Data Vault methodology and performance issues in DWH on the open-source stack. We will compare technical implementation options for the Business Vault model layer, considering the specifics of PostgreSQL and Greenplum.

We will also cover 5 real problems that arise when operating a DWH and their solutions:

1. Transferring Business Vault object assembly logic from PostgreSQL to Greenplum.
2. Slower ETL performance when building the current state of Business Vault in PostgreSQL.
3. Slower Data Lineage construction in PostgreSQL and Greenplum.
4. Slow satellite queries in Greenplum.
5. Slow queries with "IN" or "OR" in the Business Vault layer.