PG BootCamp Russia 2025 Ekaterinburg welcome speech

Michael Goldberg

PostgreSQL contributor, member of PostgreSQL Code of Conduct committee, Chair (2022-2023), member of PG BootCamp initiative group, PGDay Israel conferences organizer (as of 2017)

Vadim Yatsenko

CEO, Tantor Labs, Member of PG BootCamp Initiative Group, PGDay Israel Conferences Organizer (since 2017)

10:00 – 10:10

Guests registration

09:00 – 10:00

Conference program

Time zone: Ekaterinburg (GMT +05:00)

Implementing ILM in PostgreSQL

This talk is devoted to the implementation of the ILM (Information Lifecycle Management) approach in PostgreSQL, aimed at optimizing data storage costs through automated management of their lifecycle. We'll discuss the key aspects of ILM from basic principles, such as classifying data by demand using temperature maps, to practical mechanisms for implementing ILM policies in PostgreSQL, with special attention to setting up rules for transferring data between storage layers (for example, from SSD to HDD or object storage). We'll also figure out how the effective use of ILM can reduce infrastructure costs while preserving performance for hot data and automating the management of cold data in accordance with specified business requirements.

Igor Melnikov

Independent expert

13:10 – 14:00

Intermediate

Dinner

14:00 – 15:00

New life of the PipelineDB extension: working with time series in PostgreSQL

Modernizing the defunct PostgreSQL extension for time series aggregation is the focus of this session. I'll describe how its antiquated code was modified for use with more recent PostgreSQL versions, what improvements were made for aggregation algorithms, and how we are going to continue developing this extension which is available and free to use. I believe that developers working with time series and others interested in optimizing older systems will find my talk useful.

Vadim Iatsenko

CEO, Tantor Labs, Member of PG BootCamp Initiative Group, PGDay Israel Conferences Organizer (since 2017)

12:20 – 13:10

Coffee break

11:50 – 12:20

WAL-G: anatomy of catch up technology

I will talk about Postgres backup and some unusual features of WAL-G, one of which I will dwell on in more detail: we'll examine the internals of quickly reducing the lag between Primary and Standby, and I will give an idea of ​​​​the challenges in developing this feature now. I think this might be useful for administrators, as well as Go developers willing to get involved in open source development.

Andrey Borodin

PostgreSQL contributor, head of open-source DBMS development unit, Yandex Cloud

11:00 – 11:50

Hacking the query planner, again... and again (pt.1. Initial query transformations)

In this talk, from a developer’s view we will focus on the first phase of query optimization that is known as preprocessing of initial query structure. Listeners will know which normalization steps and one-way transformation rules are applied to incoming query by query optimizer under the hood before starting cost-based join enumeration process. In particular, we will touch topics like transforming ANY/EXISTS/IN subqueries into joins, flattening sub-selects, preprocessing expressions, reducing outer joins to inner one, pushing table predicates down to rels, removing useless joins and so on. For DBA and SQL developers we will try to expose some useful insights and hacks to make query more managed and performant. This talk is based on and greatly overlaps with two last talks on similar topic: Tom Lane's 2011 talk "Hacking the Query Planner” and "Hacking the Query Planner, Again” from Richard Guo at 2020. But here we are going to dig into the planner in more detail with closer look in source code implementation, providing at the end of the talk the transformtion patch that is able to improve the performance of some sort of queries by multiple times.

Developer, Tantor Labs

Sergey Soloviev

Maxim Milyutin

Head of R&D,Tantor Labs

10:10 – 11:00

Advanced

Advanced

Advanced

Time zone: Ekaterinburg (GMT +05:00)

Data archeology in PostgreSQL. How to “excavate” data from unreadable tables?

When PostgreSQL can no longer see a table and backups are just bitter memories, it's time for hardcore data archeology. For those who are prepared to delve further, this talk offers genuine techniques instead of illusions about "magic buttons." These include manually parsing table page numbers, looking for duplicate IDs, examining MVCC data, attempting resuscitation via pg_filedump, and dancing with WAL logs. You'll learn how to turn understanding the internals of PostgreSQL tables into a lifesaver when the clock is ticking and your business is frozen in panic. For those who don't mind getting their hands filthy with bits and bytes, there is always a genuine opportunity, even though there are no assurances or standard recovery techniques.

Mikhail Torgovkin

Senior DBMS expert, CDDT Maximum

18:10 – 19:00

Buffet

19:00 - 21:00

Eyedrops for Postgres, or how to increase the granularity of statistics without affecting DST

Standing at the crossroads of choosing the optimal query plan, PostgreSQL checks the statistics 'map'. But what if even the advanced statistics are not detailed enough, are too large in scale, or are inaccurate due to low base? In this talk, we will take a closer look at PostgreSQL statistics: what they depend on and how the PostgreSQL approach differs from ones implemented in other DBMS. Also, we will introduce a new GUC parameter and analyze its impact on the result, highlighting the conditions of applicability and assessing the pros and cons of other popular approaches.

Artem Bugaenko

Developer, Tantor Labs

17:20 – 18:10

Break

16:50 – 17:20

Upgrading PostgreSQL and Greenplum сatalogs

I will talk about binary compatibility between major versions of PostgreSQL and Greenplum, how it affects feature development and what methods are used to change the kernel or directory version.

Kirill Reshke

Developer, Yandex

15:55 – 16:50

Peculiarities in determining selectivity and related query execution issues

One of the primary factors influencing the query planner's ability to provide efficient plans for their implementation is selectivity. Determining selectivity may be challenging in some situations, and the planner may make mistakes that result in the selection of an inefficient plan. The report's theoretical section will examine the PostgreSQL query planner's selectivity determination capabilities, while in the practical part we will focus on an intriguing selectivity issue and possible debug-mode solutions.

Maxim Starkov

Senior developer, Tantor Labs

15:00 – 15:55

Advanced

Advanced

Intermediate

Intermediate

Xid, debugging, two smoking bugfixes of the 64-bit transaction counter

64-bit transaction counter (xid64) is one of the most important patches for high-load systems. Being developed in a separate thread, the patch is advancing in parallel with the Postgres itself with a constant developers' attention for compatibility. We will look at the structure and basic ideas of xid64, as well as some additional options implemented in this patch. In the practical part, we will explore some issues arising when developing and adapting the patch to newer versions. Also, we will investigate some attention-worthy errors, review the reasons for their occurrence and fix them.

Evgeny Voropaev

Senior developer, Tantor Labs

13:10 – 14:00

Dinner

14:00 - 15:00

Turning PostgreSQL into a RESTful API server

I will explain how to turn PostgreSQL into a RESTful API server, how to bidirectionally connect PostgreSQL to other databases in two clicks, how to streamline data exchange between databases of different types and quit using FWD.

Konstantin Vashchenkov

CTO, X-Square

12:20 – 13:10

Coffee break

11:50 – 12:20

Working with queries from a DBA standpoint

Using the perspective of a database administrator’s daily work, this session focuses on useful techniques for evaluating and improving SQL queries. From tools that translate parameterized queries into intelligible form to performance report analysis, we’ll examine tools that facilitate diagnostics. We will also figure out how to select indexes, configure statistics collection (the impact of the default_statistic_target parameter) and avoid typical problems: from the non-obvious consequences of using OR to the costs of DISTINCT + JOIN combinations. We’ll demonstrate how to decompose complex queries, control the optimizer using from_collapse_limit and join_collapse_limit, and evaluate the impact of parameters like n_distinct on execution time. Moreover, we will discuss query equivalence checks and limits of what DBAs can do, emphasizing that even administrators can’t always do what they want.

Alexander Nikitin

Lead DBA, DBA.Team

11:00 – 11:50

Approaches to implementing data compression

I will talk about what can be compressed in vanilla PostgreSQL now, what are the most promising compression algorithms, how compression affects TPS, and will also present our team’s approach to expanding compression, which in certain scenarios can increase performance or reduce hard drive requirements.

Pavel Seleznev

Community manager, Pangolin

10:10 – 11:00

Time zone: Ekaterinburg (GMT +05:00)

Intermediate

Intermediate

Basic

Basic

Checking PostgreSQL databases for hidden flaws with static analysis

Static analysis of the PostgreSQL database structure may be performed even on empty databases and can reveal a variety of possible issues and antipatterns, such as duplicate indexes and tables without PK or mismatched types for FK. We'll examine how to automate the search for these issues using SQL queries to the system catalog, how ORM and legacy code can cause them, how they can occasionally arise as a result of the project's rapid growth, and how to put checks in place in CI/CD to stop these "man-made" mistakes. I'll talk about practical cases and provide examples; architects, database administrators, and anyone annoyed of making "after the fact" changes to database schemas will find this content helpful.

Vasiliy Nikonov

CEO, Intellectual systems LLC

18:10 – 19:00

Фуршет

19:00 - 21:00

Changing table structure in PostgreSQL on the go using the dbms_redefinition technology

The Oracle DBMS has a package that reduces downtime to a minimum (near zero) while applying application updates that change the tables structure. This is something that is not yet available in the PostgreSQL world. In this talk, the author will introduce the dbms_redefinition extension made to implement this capability. If we take the pg_repack extension, it does NOT allow changing the table structure. So dbms_redefinition might be of interest to all PostgreSQL users regardless of whether they are migrating from an Oracle DBMS or not.

Igor Melnikov

Independent expert

17:20 – 18:10

Break

16:50 – 17:20

pg_orchestrator: testing the PostgreSQL upgrades

pg_orchestrator is a new open source utility designed to automate complex PostgreSQL update and testing scenarios. The extension allows you to describe the sequence of upgrade steps (using pg_upgrade, minor updates and pg_dumpall) using clear YAML manifests and set up test cases for performance analysis via Docker containers, not only optimizing processes, but also significantly reducing the risk of errors during migrations and queries testing. In my talk, I will tell you how this utility can help database administrators, DevOps teams and developers save time and increase system reliability. Well examine the Upgrade and Performance modules as well as real examples of configurations and scenarios.

Lev Nikolaev

Developer, Tantor Labs

15:55 – 16:50

Working with logical replication

Although logical replication is widely used for years, recently it gained some new improvements capable to significantly simplify things. This talk won’t provide any complicated theory; instead, we will focus on practical opportunities that sometimes go unnoticed. We will briefly overview the main types of replication and the limitations of logical approach, then we will proceed to basic operations with subscriptions and tables replication, to methods for resolving collisions, to creating a logical replication based on a physical one, with subsequent role switching, as well as to converting a physical replica into a logical one using various approaches. The talk is aimed at systematizing knowledge and understanding of logical replication, and highlighting related features helping save time and reduce risks.

Alexander Nikitin

Lead DBA, DBA.Team

15:00 – 15:55

Intermediate

Intermediate

Basic

Basic