Skip to content

MySQL Contributor Summit 2026 ‐ Talk proposals

Jump to bottom
Scott Stroz edited this page Jun 4, 2026 · 20 revisions

This page contains the proposal submissions received so far for the MySQL Contributor Summit. The goal is to review the proposed topics, identify possible collaboration opportunities, align related discussions, and help shape the overall structure and focus areas of the summit.

Category: AI & Cloud

Oracle:

  • Title: Native VECTOR Store and VECTOR INDEX Support for MySQL
  • URL: https://bugs.mysql.com/bug.php?id=120425
  • Abstract: This proposal aims to evolve MySQL VECTOR support from write-only metadata into a fully functional native vector store with indexing and similarity search. The goal is to support ACID-compliant AI retrieval workloads directly in MySQL, reducing the need for an external vector database.
  • Short Description of the Feature: Add native VECTOR INDEX support, vector distance/similarity functions, and ANN search capabilities so MySQL can store, query, and retrieve high-dimensional vector data efficiently within the database.

Oracle:

  • Title: Official MySQL-Native MCP Interface for AI Assistants and LLM Tools
  • URL: https://bugs.mysql.com/bug.php?id=120448
  • Abstract: This proposal explores adding an official MySQL-native MCP (Model Context Protocol) interface to enable secure and standardized interaction between MySQL and AI assistants or LLM-based tools. The goal is to provide safe schema discovery, metadata inspection, controlled query execution, and operational diagnostics directly through MySQL.
  • Short Description of the Feature: Introduce an optional first-party MCP interface/component for MySQL that allows AI tools to securely access schemas, metadata, read-only queries, and selected server diagnostics while leveraging native MySQL authentication, privilege enforcement, and operational visibility.

Category: Developer / DBA Experience

Recommended bug/feature submitted by Sveta Smirnova:

  • Title: Improve mysqlsh util.dumpInstance() chunking for composite PRIMARY KEY tables
  • URL: https://bugs.mysql.com/bug.php?id=119222
  • Abstract: This proposal addresses inefficient chunking in mysqlsh util.dumpInstance() when dumping tables with composite PRIMARY KEYs. By using the full primary key in range conditions, dump chunks can be better balanced, improving throughput and avoiding performance degradation during dump and restore.
  • Short Description of the Feature: Enhance MySQL Shell dump chunking so that tables with composite PRIMARY KEYs are chunked using the full key tuple instead of only the leading column(s). This should produce more even data distribution across chunks and improve dumping performance. The MySQL Community Developer Guide says feature requests should be filed as S4 FEATURE_REQUEST, and roadmap-aligned requests can use the ROADMAP_CANDIDATE tag; the roadmap themes include Performance and Ecosystem/tooling/connectors. For this issue, Performance is the best fit.

Readyset:

  • Title: System-versioned tables for MySQL (SQL:2011 temporal tables)
  • URL: https://bugs.mysql.com/bug.php?id=120432
  • Abstract: System-versioned tables are the SQL:2011 answer to "what did this row look like last Tuesday." Oracle, SQL Server, DB2, and MariaDB ship them today; this talk proposes bringing them to MySQL with the syntax, storage model, and design tradeoffs already validated by MariaDB's seven-year implementation, and uses the summit slot to surface the specific open design questions on which contributor input will most shape the final feature. The full Design Proposal is attached to the bug report.
  • Short Description of the Feature: Add SQL:2011 system-versioned tables to MySQL. Per-table opt-in via WITH SYSTEM VERSIONING. The server automatically maintains row_start and row_end timestamps on every DML, so SELECT ... FOR SYSTEM_TIME AS OF returns the row as it was at any past point within the retention window. This closes the largest standing SQL-standard data-management gap between MySQL and Oracle, SQL Server, DB2, and MariaDB, and replaces the trigger-driven audit-table patterns that MySQL users maintain today.

Oracle:

  • Title: DB Modeling for MySQL Workbench 10
  • URL: https://bugs.mysql.com/120481
  • Abstract: MySQL Workbench 10 is the all-new successor to MySQL Workbench, based on Electron and with support for MySQL 8.4 and newer. This proposal introduces updated ER-Modeling capabilities to the new MySQL Workbench.
  • Short Description of the Feature: Add visual database modeling capabilities supporting the full current MySQL feature set, improved reverse and forward engineering, synchronization, Git integration and more.

Oracle:

  • Title: Native Support for BOOLEAN, UUID, and SQL ARRAY Data Types in MySQL
  • URL: https://bugs.mysql.com/bug.php?id=120450
  • Abstract: This proposal explores adding first-class native support for commonly requested data types such as BOOLEAN, UUID, and SQL-standard ARRAY types in MySQL. The goal is to improve standards compliance, simplify application development, provide clearer metadata semantics, and reduce reliance on workarounds based on aliases, functions, JSON, or application conventions.
  • Short Description of the Feature: Introduce native implementations of BOOLEAN, UUID, and potentially SQL ARRAY data types with standardized behavior, metadata support, validation, storage optimization, indexing, replication semantics, and improved interoperability with connectors and developer tools.

Oracle:

  • Title: Native DATETIME WITH TIME ZONE Support in MySQL
  • URL: https://bugs.mysql.com/bug.php?id=120454
  • Abstract: This proposal introduces a native DATETIME WITH TIME ZONE data type for MySQL to provide standard-compliant temporal handling beyond the limitations of the current TIMESTAMP implementation. The work also includes a broader refactoring of MySQL temporal type internals to improve consistency, type safety, and future extensibility.
  • Short Description of the Feature: Implement a native DATETIME WITH TIME ZONE temporal type that stores UTC values together with associated time-zone displacement information, while refactoring MySQL’s internal handling of DATE, TIME, DATETIME, and TIMESTAMP types to improve correctness, maintainability, and support for future temporal enhancements.

Percona:

  • Title: Provide a better way for chunking a dump of tables with composite primary key in MySQL Shell
  • URL: https://bugs.mysql.com/bug.php?id=119222
  • Abstract: Creating a logical dump of large tables may take a long time, thus can lead to performance degradation of application queries running in parallel. Therefore it is very important to create such dumps as fast as possible. Our feature proposal uses all columns in the composite primary key to find a way to dump data in parallel instead of relying on the first integer column like it is currently implemented in MySQL Shell.
  • Short Description of the Feature: Better dump performance for large tables with composite primary keys.

Category: Extensibility / Ecosystem / Tooling

MariaDB:

  • Title: MariaDB Foundation as Part of the MySQL Ecosystem
  • URL: N/A
  • Abstract: The short talk would focus on how MariaDB Foundation views itself as a constructive participant in the broader MySQL ecosystem, how we see interoperability and extensibility evolving, and how some MariaDB innovations may contribute useful perspectives to the broader ecosystem discussion.
  • Short Description of the Feature:

Oracle:

  • Title: Inside MySQL’s Component Framework: Building Extensible Features the MySQL Way
  • URL: N/A
  • Abstract: MySQL’s component framework is one of the key foundations for extending the server functionality. This talk introduces how components are structured, how services are registered and consumed, and how component-based features fit into the MySQL server architecture.
  • Short Description of the Feature: TBD

VillageSQL:

  • Title: Toward a Stable, Versioned Extension API and ABI for MySQL
  • URL: https://bugs.mysql.com/bug.php?id=120427
  • Abstract: This session explores the challenges of extending MySQL using today’s plugin and component infrastructure, where API and ABI compatibility across versions and distributions remains difficult to guarantee. It introduces the VillageSQL Extension Framework (VEF) and a proposed shared extension API/ABI model aimed at improving portability, stability, and long-term extensibility for the MySQL ecosystem.
  • Short Description of the Feature: Current MySQL extensibility mechanisms lack a stable, versioned API/ABI model, making it difficult to build portable extensions that work reliably across MySQL versions and compatible distributions. This proposal suggests introducing a shared extension API/ABI specification with standardized extension lifecycle management.

ACE/RS Alkin T.:

  • Title: MyVector: Component Architecture Migration for MySQL 8.4 LTS and 9.7 LTS
  • URL: https://bugs.mysql.com/bug.php?id=120433
  • Abstract: MyVector brings native vector similarity search to MySQL as a community extension, and its latest release (v1.26.5) marks a pivotal step: migrating from the legacy plugin model to the MySQL Server Component architecture on MySQL 8.4 LTS and 9.7 LTS. This presentation covers the component build design, the key correctness fixes uncovered during the migration (binlog listener crash loop, KNNIndex double-insert on reconnect, stale checkpoint bug), and the roadmap for graduating the component path to the default deployment model.
  • Short Description of the Feature: MyVector is an open-source native vector similarity search plugin for MySQL (GPL-2.0) that implements HNSW and KNN approximate nearest-neighbor search via SQL UDFs and stored procedures. Version 1.26.5 introduces a MySQL Server Component build path (INSTALL COMPONENT) as the forward-looking deployment model for MySQL 8.4 LTS and 9.7 LTS, alongside the existing plugin path, which remains stable for MySQL 8.0 through EOL.

Contributor: Daniel V.E.:

  • Title: MySQL network protocols
  • URL: N/A
  • Abstract: This proposal focuses on the future direction of MySQL network protocols, including the relationship between the Classic Protocol and X Protocol, ecosystem adoption, connector support, and protocol documentation. The discussion aims to explore whether the Classic Protocol should continue evolving, clarify the long-term role of X Protocol, and identify improvements needed to better support third-party connectors, proxies, routers, and broader ecosystem interoperability.
  • Short Description of the Feature: Evaluate the future direction of MySQL network protocols, including enhancements to the Classic Protocol, the long-term role of X Protocol, improved ecosystem interoperability, connector adoption, and modernization of protocol documentation for third-party tools and implementations.

Amazon:

  • Extending the MySQL Component Framework into the Query Engine
  • URL: N/A
  • Abstract: MySQL's component framework has matured into the project's preferred extensibility surface — audit, keyring, validate_password, and telemetry have all migrated from plugins to components. However, the component surface does not yet reach the query engine: SQL data types, index access methods, SQL operators, and optimizer hooks remain outside the framework. This talk proposes growing the component framework into the query engine in phases — operator registry and optimizer hooks first (additive, no protocol impact), followed by type registry and index access method service. VECTOR is proposed as the first use case, requiring all three surfaces. The proposal is complementary to VillageSQL's Stable Extension API/ABI talk and builds on Oracle's existing component infrastructure rather than introducing a competing model.
  • Short Description of the Feature: Grow MySQL's component framework into the query engine by opening SQL data types, index access methods, SQL operators, and optimizer hooks as component services. VECTOR is the first use case. The goal is to make MySQL a platform where Oracle's own roadmap items — VECTOR, BOOLEAN, UUID, ARRAY — are built using the same framework available to all contributors.

Readyset:

  • Title: Support InnoDB-based binary log (transactional binlog stored within InnoDB)
  • URL: https://bugs.mysql.com/bug.php?id=120430
  • Abstract: This session proposes an InnoDB-based transactional binary log architecture for MySQL, where binlog events are stored directly within InnoDB instead of standalone files. The approach aims to simplify durability and crash recovery, reduce commit overhead by eliminating an extra fsync and two-phase commit coordination, and provide transactionally consistent binlog state for replication and CDC ecosystems, while preserving the existing replication protocol.
  • Short Description of the Feature: Move the binary log out of standalone files and into an InnoDB-managed tablespace, so that data changes and their binlog events are written within the same mini-transaction and made durable by a single redo flush. This removes one fsync from every durable commit, eliminates the two-phase commit between InnoDB redo and the binary log, simplifies crash recovery to a single redo pass, and makes binlog state transactionally consistent with the rows it describes. The replication wire protocol stays unchanged. The feature is opt-in via a new binlog_storage = FILE | INNODB setting, defaulting to FILE.

Category: Performance & Observability

Oracle:

  • Title: DDL Triggers for Schema Change Auditing and Control in MySQL
  • URL: https://bugs.mysql.com/120435
  • Abstract: This proposal introduces native DDL triggers in MySQL to allow auditing, validation, and control of schema-level changes such as CREATE, ALTER, DROP, and RENAME operations. The feature would provide administrators and application teams with built-in mechanisms for governance, compliance, operational safety, and schema change visibility.
  • Short Description of the Feature: Add support for BEFORE and AFTER DDL triggers that execute automatically on schema/object modification events, enabling native auditing of schema changes, enforcement of operational policies, and prevention of risky or unauthorized DDL operations directly within MySQL.

Contributor: Zhao Song:

  • Title: InnoDB B+Tree Performance Optimization, Part 1: Insert Path Improvements and Concurrent Split Handling
  • URL: https://bugs.mysql.com/bug.php?id=120434
  • Abstract: This talk presents a PoC redesign of InnoDB’s B+Tree insert and page split path. In a split-heavy 32-thread insert workload on MySQL 9.7.0, the PoC improved TPS by about 16x and shifted the main bottleneck from index-lock contention to redo log commit wait.
  • Short Description of the Feature: This proposal aims to improve InnoDB B+Tree insert performance under high-concurrency, split-heavy workloads. It focuses on reducing repeated B+Tree descents in the insert path and enabling more concurrent handling of page splits inspired by B-link Trees.

Oracle:

  • Title: Improving Hypergraph Optimizer Performance and Feature Parity
  • URL: https://bugs.mysql.com/bug.php?id=120452
  • Abstract: This proposal focuses on improving the MySQL hypergraph optimizer by addressing performance regressions and closing remaining functional gaps compared to the legacy optimizer. The goal is to deliver more stable and competitive query plans across a broader range of workloads without requiring users to fall back to the old optimizer.
  • Short Description of the Feature: Enhance the hypergraph optimizer through improved cardinality estimation, costing, join ordering, access path selection, query transformations, and support for additional plan alternatives to reduce performance regressions and achieve closer parity with the legacy optimizer.

Oracle:

  • Title: Structured JSON Optimizer Trace for the Hypergraph Optimizer
  • URL: https://bugs.mysql.com/bug.php?id=120453
  • Abstract: This proposal introduces a structured, machine-readable JSON optimizer trace for the MySQL hypergraph optimizer to improve debugging, visualization, and regression analysis. The goal is to expose optimizer decisions, estimates, and plan alternatives in a stable format that can be consumed by tools and advanced users.
  • Short Description of the Feature: Replace or complement unstructured hypergraph optimizer trace messages with a structured JSON-based trace format that exposes candidate plans, join ordering, access paths, cardinality estimates, cost calculations, and plan selection reasoning in a tool-friendly manner.

Oracle:

  • Title: Built-in MySQL Health Monitor for On-Premises Deployments
  • URL: https://bugs.mysql.com/bug.php?id=120447
  • Abstract: This proposal introduces a built-in MySQL Health Monitor for on-premises environments to improve operational visibility, diagnostics, and proactive resource monitoring directly from the server. The feature would expose critical health metrics through SQL-accessible interfaces and help DBAs detect resource pressure before it impacts stability.
  • Short Description of the Feature: Provide a native server-side Health Monitor that exposes system health information such as disk usage, memory pressure, CPU activity, process activity, and resource trends through Performance Schema, including warnings for critical operational conditions.

Oracle:

  • Title: Partition Pruning Visibility in Iterator-Based EXPLAIN Output
  • URL: https://bugs.mysql.com/bug.php?id=120451
  • Abstract: This proposal improves EXPLAIN consistency in MySQL by exposing partition pruning information in tree-based and iterator-based EXPLAIN formats, including EXPLAIN ANALYZE and JSON format version 2. The goal is to provide developers and DBAs with clearer visibility into partition selection behavior across all query plan representations.
  • Short Description of the Feature: Extend iterator-based EXPLAIN output formats to display selected partition information on relevant table access nodes, ensuring consistent partition pruning visibility across traditional, tree-based, hypergraph, and JSON EXPLAIN outputs.

Oracle:

  • Title: Controlled Deletion of Obsolete UUIDs from GTID State
  • URL: https://bugs.mysql.com/120460
  • Abstract: This proposal introduces a safe, online mechanism for retiring obsolete server UUIDs from MySQL GTID state without disrupting replication correctness or GTID-based tooling. The goal is to reduce GTID metadata bloat and operational complexity in environments where servers are frequently replaced, cloned, or decommissioned.
  • Short Description of the Feature: Add an explicit, privilege-protected, staged process for marking obsolete UUIDs for deletion, propagating that removal state through the topology, and completing removal only when it is safe. The feature should preserve replication, backup, restore, failover, and consistency semantics while allowing administrators to clean up stale GTID history.

Amazon:

  • Title: Performance Improvements and Vector Support

  • URL: https://bugs.mysql.com/bug.php?id=120407 & https://bugs.mysql.com/bug.php?id=120408

  • Abstract: This presentation highlights key gaps in MySQL impacting performance, developer efficiency, and emerging workloads. We discuss three practical improvements—RETURNING support, a JOIN regression linked to Adaptive Hash Index behavior, and io_uring integration—alongside an evaluation of the new VECTOR data type, which currently lacks essential query capabilities such as distance functions and indexing. Based on reproducible tests and community feedback, we outline concrete directions to strengthen MySQL’s competitiveness in both traditional and AI-driven use cases.

  • Short Description of the Feature: 120407: This proposal targets three high-impact gaps in MySQL affecting performance and developer efficiency: missing RETURNING support leading to unnecessary round-trips and prolonged locking, a persistent JOIN regression linked to Adaptive Hash Index behavior, and lack of io_uring integration for modern high-throughput I/O. We outline reproducible cases, community evidence, and concrete implementation directions to address these limitations. 120408: This proposal evaluates the current state of MySQL’s VECTOR data type in 9.7 Early Access, highlighting critical gaps that limit it to storage-only use. While basic vector storage and conversion functions are functional, the absence of distance functions, vector indexes, and similarity search prevents practical adoption. We compare MySQL with MariaDB and PostgreSQL (pgvector), outline missing capabilities, and discuss concrete steps to enable full vector search functionality.

  • Title: InnoDB Batch Commit Support for Optimized Group Commit and Replication Apply

  • URL: https://bugs.mysql.com/120461

  • Abstract: This proposal introduces a native batch commit capability in InnoDB to allow coordinated commits of transaction groups as a single storage-engine operation. The goal is to improve source-side group commit efficiency and provide a stronger foundation for optimized replication apply on replicas.

  • Short Description of the Feature: Add a batch commit interface to InnoDB and the handlerton layer that enables coordinated atomic visibility and ordered persistence of grouped transactions. The feature is intended to reduce commit overhead, improve binary log group commit efficiency, optimize replica apply workloads, and provide more scalable transaction processing under high concurrency.

Oracle:

  • Title: Next-Generation MySQL Replication Performance and Scalability Improvements
  • URL: https://bugs.mysql.com/120462
  • Abstract: This proposal explores a broad set of improvements aimed at increasing MySQL replication throughput, reducing latency, improving recovery efficiency, and simplifying replication operations at scale. The discussion focuses on modernizing replication infrastructure, optimizing commit processing, enhancing CDC capabilities, and improving both source and replica performance under demanding production workloads.
  • Short Description of the Feature: Investigate and prioritize replication enhancements including streaming replication, optimized batch commit processing, Binlog Group Commit improvements, modernization of binary log infrastructure, optional relay log models, enhanced CDC frameworks, richer binlog metadata, and more flexible replication filtering to improve scalability, operational efficiency, and recovery performance.

Clone this wiki locally