FOSDEM 2026 - Notizen
/ home / computer / .
preFOSDEM MySQL Belgian Days 2026
Agenda
The PlanetScale Binlog Server
Shlomi Noach, PlanetScale/Vitesse
- Inventor Jean-François Gagné
- Duplicate Binary Log (possible to store to “slow” hdd?, S3)
- Works semi-sync
- Serve binary logs
- Only one singe replication tree
- See also Percona binlog server
Binary Log Server - the missing MySQL infrastructure component
Yura Sorokin, Percona, Source
mysqlbinlog --from-remote-server is very limited
- Blackhole SE: Overhead in Level 0 slaves → opportunity:
- Percona Binlog Server
- Much less overhead compared to
mysqld
- Works in 2 modes:
- Fetch: until last or error
- Pull: continues + reconnect
InnoDB: Deep dive into Dynamic Redo logging and Instant DDL
Dimitri Kravtchuk
- Dynamic REDO configuration
- Tail: Checkpoint-age
- InnoDB log streaming
*_tmp: not in user
- 32 log files
- See also here and here
- Details you will find in
P_S
-
- Instant ADD/DROP COLUMN
- → Version → check in
P_S
- Best practice:
ALGORITH = INSTANT
- See also
I_S.innodb_tables
- Page fragmentation hints
Beyond Linear Read-Ahead: Logical Prefetching using Primary and Secondary Indexes in InnoDB
Vitor Oliveira (Huawei)
Innovating in MySQL with Extensions
Tomas Ulin and Dominic Preuss
- Tomas Ulin → own company: VillageSQL
- A branch of MySQL
Heather VanCura, Oracle
Bringing Flamegraphs to MySQL Query Optimization
Vinicius Grippa, Percona
EXPLAIN ANALYZE with: explain_json_format_version = 2
myflames on Github
FOSDEM, Saturday
Agenda
PostgreSQL and MySQL, Two Databases, Three Perspectives
Rohit Nayak, Shlomi Noach, Ben Dicken, Pep Pla
- PostgreSQL
- License and Community (+)
- Extensions (+)
- Vibes (+)
- Architecture
- Clustered table vs. heap table
- Processes vs. threads → Pooling necessary if > 100 connections,
pg_bouncer, ProxySQL (- PostgreSQL)
- Index flexibility (+ PostgreSQL)
- Storage Engine →
pg_repack, pq_squeeze, OriolDB
- Extensibility (+ PostgreSQL)
- Replication (- PostgreSQL)
- MVCC
- PostgreSQL: in table
- InnoDB: in undo
- In PostgreSQL reads can provoke writes (visibility bit)!
- InnoDB: No bloat
- Clean-up:
- InnoDB page cleaner process
- PostgreSQL: VACUUM
- Logical replication:
- MySQL logical binlog application
- PostgreSQL → WAL, physical copy
- MySQL: all changes
- PostgreSQL: only row changes + relation info
“Drop-in Replacement”: Defining Compatibility for Postgres and MySQL Derivatives
Jimmy Angelakos, Daniël van Eeden (pgEdge)
Jack of all trades: query federation in modern OLAP databases
Nicoleta Lazar
+--------------+
| ... |
+--------+ +-------+ +--------------+
| Client | -> | Trino | -> | Data Sources |
+--------+ +-------+ +--------------+
| ... |
+--------------+
+--------------+
+---------------------------+ | ... |
+--------+ |+-----------+ +-----------+| +--------------+
| Client | -> || StarRocks | | DataCache || -> | Data Sources |
+--------+ |+-----------+ +-----------+| +--------------+
+---------------------------+ | ... |
+--------------+
Cracking Down the Code: What Really Happens When You Run a SELECT?
Charly Batista
- SQL - The wish
- Planer - The bet
- Executor - The bill
- Parser → Tree profile
- Analyzer → Query tree
- Rewriter
- Planer + Statistics
- Executor - Measure shared buffer usage
Vitor Oliveira (Huawei)
Delegating SQL Parsing to PostgreSQL
Greg Potter
*.sql → empty DB → query DB → system catalog + pg_class, pg_depend, pg_attribute
- Stores the graph
Replicating Transactional Databases to ClickHouse: Transaction Log Analysis and Time Travel
Arnaud Adant
MariaDB Day – Brussels
Agenda
Meet 12.3 LTS
Nikita Malyavin, MariaDB Corp.
New features in MariaDB 12.3 LTS
SET SESSION AUTHORIZATION user@host
- File key management
- SQL:
IS JSON
SET PATH
- Storage Engine:
VIDEX (hypothetic indexes) → EXPLAIN
- Performance
aria_page_cache segments, for temporary tables
- MDL lock sharding
- Replication
wsrep_applier_retry_count
slave_parallel_threads
- Storing binary log in InnoDB!!! → Slides
- Oracle compatibility
How MariaDB Achieved 2.5× OLTP Throughput in Enterprise Server 11.8
Steve Shaw, HammerDB (“leads” the MariaDB Plc. performance/bencharking team!!!)
MariaDB is the Future of MySQL
Michael “Monty” Widenius, MariaDB Corp
- OSS vs. commercial
- PostgreSQL
- Connections/threading
- Logical replication
MySQL 8.0 Galera Cluster to MariaDB Galera Cluster migration: Theory and Practice
Jan Lindström, MariaDB Corp
Peter Zaitsev, Percona
Seamless Schema Changes: Mastering Non-Blocking ALTER TABLE in MariaDB
Nikita Malyavin, MariaDB Corp
Rebuildilng Code with CodeWolf and Mariadb Galera Cluster with Vector Databases
Paul Clevett, Wolf Software Systems Ltd
Optimising Ingres Data Migration
Nick Deninng, Diegesis
Deploying RAG Without the Plumbing: MariaDB’s Unified Approach
Alejandro Duarte, MariaDB Corp.
- RAG: Context is all you need
- MariaDB AI RAG → Enterprise
- Model LLM für Laptop für RAG
- See video on YouTube
The blissful boredom of choosing MariaDB
Dirk Hillbrecht, Cantamen
(Almost) No Downtime MariaDB Upgrades
Andrija Vučinić, Automattic
Building the European Cloud with MariaDB
Carl Schwan, Nextcloud
Running MariaDB When Nobody Wants to Be the DBA: From Manual Guesswork to Safe Automation
Roman Agabekov, Releem
HyperDB
- On Github
- Query Response time plug-in
Announcement
Some statements noted at the Summit organized by Percona:
- MySQL is becoming a DB2.
- PostgreSQL is still a bitch to operate.
- There are significant differences between user behaviour in Europa/North America and Asia.
- 50% of MySQL staff was RIFfed (Reduction in Force = fired) recently at Oracle. But after significant growth in the last years. We are talking about 50 Developers now.
- MySQL is at risk at #Oracle.
- MySQL is not dead but in “maintenance mode”.
- #Alibaba has more developers working on MySQL than Oracle.
- Why PostgreSQL is winning: Extensions!
Companies seen at this Summit: Percona, MariaDB, Oracle, AWS, Ubuntu/Canonical, WordPress, booking.com, ProxySQL, and many other huge players of the ecosystem.
Summary by Kaj Arnö, MariaDB Foundation? having stakes in the MariaDB Corp, 2026-02-03 : Reading the Room: What Europe’s MySQL Community Is Really Saying
Why PostgreSQL is winning? → Extentsions! I am not so sure if this technical explanation is the only one…