CERN PGDay 2026

Friday Feb 6, 2026, 9:30 AM → 9:30 PM Europe/Zurich

503/1-001 - Council Chamber (CERN)

After a successful first edition in 2025, CERN PGDay returns in 2026 as a regular gathering for PostgreSQL users and enthusiasts in Suisse Romande (western Switzerland). Co-organized by CERN and SwissPUG, the event offers a chance to connect, share ideas, and exchange experiences in the vibrant Geneva region — home to many international organizations across the public, private, and scientific sectors.

The 2026 edition will take place on Friday, February 6th, starting at 10:00. The programme features a single track of seven sessions, all held in English, followed by a social event to continue discussions and networking in the inspiring setting of CERN. 

Registration

Registration is open until 2nd February 2026 23:59 CET and operated via the event page directly at CERN. This allows the automatic issuing of the CERN visitor badges that are needed to enter the campus. Members of SwissPUG and CERN staff or alumni are eligible for a discounted rate. Payments can be made by credit card or bank transfer.

 

Sponsors

 

Partners

Call for Sponsors

We are offering five equal sponsorship opportunities of CHF 1'000 each. These are available on a first-come, first-served basis. Please contact the organising team for details.

How to get to and visit CERN

CERN is an exciting place that can be reached in about 25 minutes from Switzerland via Geneva and tram 18. There are also buses (68 & 67) if you are coming from France. For more details on directions click here.

Have a look at your schedule before or after the event: you might be able to take advantage of the conference to visit.cern

At the site

For Internet access, the public WiFi network "CERN-Visitors" or eduroam are available. More information on campus access and directions to the event location on the CERN campus can be found here.

Organization Team

The organization is mostly done by a local team, with support from SwissPUG. The team consists of:

• Maurizio De Giorgi, CERN
• Andrzej Nowicki, CERN
• Tobias Bussmann, SCNAT
• Andreas Geppert, ZKB
• Markus Wanner, EDB
• Daniel Westermann, dbi

Call for Papers

We are keen to get proposals for talks via Indico. The call for Papers will be open until 7th December 2025.

This year's talk selection committee consists of:

• Tobias Bussmann, SCNAT (chair, non voting)
• Abel Cabezas Alonso, CERN
• Maurizio De Giorgi, CERN
• Andreas Geppert, ZKB
• Julia Gugel, migrolino
• Tom Hagel, Volue
• Svitlana Lytvynenko, CYBERTEC
• Gülçin Yıldırım Jelinek, Xata
  
  

Each member has an equal vote and abstains from voting on his or her own proposals or those of colleagues. The talk selection committee can be reached at cern-pgday-tsc@swisspug.org. CERN, venue host and co-organizer of the event, will provide a keynote presentation on relevant topics to be agreed with the committee. We will notify all submitters of the outcome of the selection process by 19 December 2025.

Code of Conduct

PostgreSQL conferences have an amazing atmosphere and are very welcoming to new people joining the community. This is something we treasure. Therefor we require all attendees, speakers, sponsors and volunteers to follow the PostgreSQL Project's Code of Conduct during the event.

Thank you for your help in making the CERN PGDay 2026 fun and enjoyable for everyone!

If needed, please contact any organizer, or a member of the community Code of Conduct Committee.

Red Lanyards

CERN PGDay operates a "Red Lanyard" scheme that is designed to help you maintain your privacy at our events. When you collect your badge upon arrival at the conference, you will be able to select either a red or white lanyard. Choosing a red lanyard will indicate to the event organisers and other attendees that you do not wish to have your photograph taken and posted on social media or on other public sites.
We will make every effort to honour your wishes, and we ask all attendees to do the same. It is, however, a 'best effort' system and sometimes a red lanyard may not be spotted before a picture is posted publicly. In cases where this happens, please reach out to either the person who posted the picture or to the conference organisers to request that the picture be removed.

Financial disclosure

The budget of CERN PGDay is calculated to break even with the support of our generous sponsors. Any deviations from the budget will be borne by the Swiss PostgreSQL Users Group and any profits will be invested in future activities to support the PostgreSQL community according to their statutes.
No CERN PGDay staff receives any payment for their work, except free attendance and a staff dinner.

The CERN PGDay is a community event organized by the Swiss PostgreSQL Users Group in collaboration with CERN.

Useful information

Local points of contact: Organization team, CERN Security Service: +41 22 76 76666

In case of emergency or incident: call +41 22 76 74444 (Tel. 74444)

"PGDay" is a registered trademark of PostgreSQL Europe. 

9:30 AM → 10:00 AM Welcome coffee

10:00 AM → 10:05 AM Opening

Welcome from the organizing team, logistics info and notices, sponsors presentation and special thanks

Speakers: Maurizio De Giorgi (CERN), Markus Wanner

pgday-cern-2026-opening.pdf

Recording

10:05 AM → 10:35 AM A new PostgreSQL backend for CERN Tape Archive scheduling for LHC Run 4

The CERN Tape Archive (CTA) stores over one exabyte of scientific data. To orchestrate storage operations (archival) and access operations (retrieval), the CTA Scheduler coordinates concurrent data movements across hundreds of tape servers, relying on a Scheduler Database (Scheduler DB) to manage the metadata of the in-flight requests. The existing objectstore-based design of the CTA Scheduler DB is a complex transactional management system. This talk presents the development of a new PostgreSQL-based backend for the CTA Scheduler as an off-the-shelf solution which simplifies implementation and is expected to significantly reduce future development and operational costs. We describe the implementation of all main CTA workflows and explain how PostgreSQL addresses the limitations of the objectstore-based system, providing the foundation for the tenfold increase in data throughput expected during LHC Run 4.

Speakers: Dr Jaroslav Guenther (CERN), Konstantina Skovola

PGDay_2026-A new PostgreSQL backend for CERN Tape Archive scheduling for LHC Run 4 - Final v4.pdf

Recording

10:35 AM → 11:05 AM DCS Data Tools - PostgreSQL/TimescaleDB Implementation for ATLAS DCS Time-Series Data

The ATLAS Detector Control System manages over 68 TB of time-series data accumulated since 2007. This presentation describes the practical implementation and operational deployment of TimescaleDB—a PostgreSQL extension—to modernize DCS data access for the ATLAS experiment. We share our experience as PostgreSQL users and administrators implementing a production time-series database solution in a large-scale scientific environment. The implementation delivers significant value through improved user experience: scientists and detector experts now access 18 years of historical data through simple Python APIs, Grafana dashboards, and C++ interfaces, eliminating the need for complex SQL knowledge. TimescaleDB's native compression achieves 11× storage reduction (68 TB to 6.6 TB), while query performance improvements enable real-time correlation studies across multiple detector systems. The PostgreSQL ecosystem enabled rapid development—Grafana integration, LDAP authentication, and CERN's Database-on-Demand (DBOD) infrastructure worked seamlessly, accelerating deployment. This PostgreSQL-based approach has unlocked new operational capabilities: detector experts built custom monitoring applications, predictive maintenance workflows identify hardware failures before they impact data collection, and commissioning teams correlate real-time measurements across detector subsystems. The system now serves production users with continuous availability and sub-minute data latency. We discuss practical deployment challenges, infrastructure lessons learned while working with CERN DBOD team, performance tuning strategies (chunk sizing, WAL management, JIT optimization), and the organizational impact of choosing PostgreSQL's open-source ecosystem for long-term scientific data management.

Speakers: Dimitrios Matakias (CERN), Paris Moschovakos

dcs_data_tools_pgday_2026.pdf

Recording

11:05 AM → 11:15 AM Break

11:15 AM → 12:00 PM Operational hazards of managing PostgreSQL DBs over 100TB

Picture this: you start a new role, eager to learn and contribute with your ideas! Your next task is to get familiar with the database setup, and then you start encountering these massive PostgreSQL databases — 100TB, 200TB, 300TB...

And you start questioning yourself: how do you backup (and restore) a +100TB database? And how about HA? Performance? Vacuum?

It should work the same way as for a 100GB database, right? Well, maybe not exactly.

Blog posts and best practice guides make PostgreSQL seem straightforward—until you push it to its limits. At extreme scale, you will find yourself questioning the most fundamental assumptions about how PostgreSQL works.

Over the last years, my team at Adyen has been exploring the boundaries of what PostgreSQL can do, and today I will share our findings with you (at least the ones I can!).

Speaker: Teresa Lopes (Adyen)

CERN-PGDAY-2026-TeresaLopes.pdf

Recording

12:00 PM → 1:15 PM Lunch break

1:15 PM → 2:00 PM Vacuuming Large Tables: How Recent Postgres Changes Further Enable Mission Critical Workloads

We have all heard about Postgres vacuum horror stories and tales of transaction wraparound disasters. Even if you've never been through one yourself, you may be concerned that you might someday experience it, or maybe you even know people who have avoided Postgres altogether due to fear of it happening to them. But it doesn't have to be this way.

In this talk, we will explore this topic through a review of a real-world wraparound incident, walking through the challenges and hope offered by changes in the last several Postgres releases. We'll explore the inner workings of the Postgres vacuum process, its dual purpose of managing both disk bloat and transaction IDs (XIDs), and how recent innovations have improved this process, particularly for large tables with heavy transaction loads. We'll dive deep into key features like: - How index de-duplication minimizes vacuum workloads and improves efficiency. - How on/off index vacuuming gives you better control during maintenance windows. - Why one key autovacuum enhancement may be the key for eliminating XID wraparounds.

Beyond theory, we'll revisit the real-world incident and see how each new Postgres feature could have helped mitigate the impact. While no single change is a silver bullet, the combined effect is significant. We'll also explore how the radix tree implementation in Postgres 17 might be the possible final piece of the puzzle for providing a manageable solution to XID wraparound risks.

This talk equips DBAs with a deeper understanding of recent Postgres advancements and their practical application in managing high-transactional and/or mission-critical workloads. You'll leave with actionable insights to optimize your Postgres environment and concrete arguments to convince management that it is worth the upgrade for the potential to minimize the risk of XID wraparound incidents.

Speaker: Robert Treat

2:00 PM → 2:10 PM Break

2:10 PM → 2:55 PM The (very practical) Postgres Sharding Landscape

Sharding enables horizontal scaling of Postgres, by logically splitting your database into several subsets ("shards"). Each shard will be a primary or, for high availability purposes, a primary-replica cluster. This allows you to scale writes horizontally, allowing Postgres to reach database sizes of multiple terabytes or even petabytes.

Sharding solutions for Postgres exist in multiple forms, such as embedded into Postgres as an extension, using native Postgres techniques and as external components, including Citus, Apache ShardingSphere, PgDog, native partitions with FDW, and others. This talk will:

• Analyze the evolving landscape of sharding solutions for Postgres.
• Offer advice and recommendations on which one to select.
• Provide open source code for simple demos so you can quickly evaluate them in your own environment.
• Present a real-world case study: planning the migration of a 1PB+ workload to sharded Postgres at a major US telco.

If demo Gods will allow, demos of sharding Postgres will be performed live.

Speaker: Álvaro Hernández

Recording

2:55 PM → 3:25 PM Coffee break

3:25 PM → 4:10 PM The Alchemy of Shared Buffers: Balancing Concurrency and Performance

Talk dissects PostgreSQL’s shared buffers: the three‑layer design (buffer pool → BufferDesc headers → buf_table hash), the hit/miss/I‑O lifecycle, and how pins, per‑page LWLocks, and atomic BM_* flags coordinate page‑granular concurrency. We will trace clock‑sweep and buffer rings that prevent big scans and VACUUM from polluting the cache, then follow WAL‑before‑data through bgwriter and the checkpointer to show how writes are made smoother. We will also discuss tuning and what’s new in PG 17 (vectored reads), and PG 18 (async reads) and what can come next - we will dive into new design proposals discussed in 2025 - to allow dynamic resize of shared_buffers via segmented shared memory.

Key takeaways
- Architecture & locks: three layers; pins + per‑page LWLocks + partitioned mapping yield high concurrency
- Replacement & pollution control: clock‑sweep keeps hot pages; buffer rings fence off bulk scans and VACUUM
- Durability & flushing: WAL‑before‑data with bgwriter/checkpointer turns random writes into managed I/O
- Tuning: balance double buffering, size shared_buffers sensibly, smooth checkpoints, and verify with EXPLAIN (BUFFERS) and pg_stat_bgwriter
- Roadmap: PG 18’s async reads cut scan latency; dynamic shared_buffers resizing is being prototyped

Speaker: Josef Machytka

postgresql_shared_buffers.pdf

Recording

4:10 PM → 4:20 PM Break

4:20 PM → 5:05 PM Incremental Backup with PostgreSQL17

In this talk, we will explore the powerful incremental backup feature introduced in PostgreSQL 17. Rather than backing up the entire database, incremental backups focus on the changes made since the last backup. This approach significantly reduces time and storage requirements compared to full backups, making incremental backups an efficient, time-saving solution for data backup processes.

The talk will begin by explaining the fundamental concepts of incremental backups. We will then analyse how to handle retention over multiple backup increments. Finally, we will examine the key benefits of incremental backups, such as reduced backup times, lower storage costs and minimized impact on the database performance.

To provide a clear and practical understanding, I will present a series of examples and tests that we have conducted. These will illustrate how to implement and utilize incremental backups, making the concept easier to grasp.

Speaker: Ms Marion Baumgartner

Incremental Backup with PostgreSQL 17

Incremental Backup with PostgreSQL 17

Recording

5:05 PM → 5:20 PM Closing

Closing notes with sponsors review and special thanks

Speakers: Maurizio De Giorgi (CERN), Markus Wanner

pgday-cern-2026-closing.pdf

Recording

5:20 PM → 7:30 PM Social event