TiDB Roadmap

This roadmap provides a look into the proposed future. This will be continually updated as we release long-term stable (LTS) versions. The purpose is to provide visibility into what is coming, so that you can more closely follow the progress, learn about the key milestones on the way, and give feedback as the development work goes on.

In the course of development, this roadmap is subject to change based on user needs and feedback. DO NOT schedule your release plans according to the content of the roadmap. If you have a feature request or want to prioritize a feature, please file an issue on GitHub.

Rolling roadmap highlights

CategoryEnd of CY24 LTS releaseMid of CY25 LTS releaseFuture releases
Scalability and Performance
Enhance horsepower
  • TiKV in-memory data cache
    TiKV maintains recent versions of data in memory to reduce redundant MVCC scans, thus improving performance.

  • Global indexing for partitioned tables

  • Adaptive concurrency for statistics collection
    TiDB automatically adjusts the parallelism and scan concurrency of statistics collection tasks based on the number of deployed nodes and hardware specifications, improving the speed of statistics collection.

  • Rapid database recovery
    Reduce the time required for full database recovery and point-in-time recovery (PITR).

  • Unlimited-size transactions
    The volume of data processed by uncommitted transactions is no longer limited by the available memory of TiDB nodes, thus improving the success rate of transactions and batch tasks.

  • Load-based traffic routing by TiProxy
    TiProxy forwards traffic based on the workload of the target TiDB node, maximizing hardware resource utilization.
  • Microservice for PD heartbeat
    Heartbeat services in PD can be independently deployed and scaled, preventing PD from becoming a bottleneck for the cluster’s resources.

  • Less I/O consumption for statistics collection
    Users can choose to scan only a portion of the data samples on TiKV during statistics collection, reducing time and resource consumption.

  • Remove the limitation for pushing down Limit operator to TiKV

  • Cascades optimizer framework
    Introduce a more mature and powerful optimizer framework to expand the capabilities of the current optimizer.

  • Single DM task reaches 150 MiB/s during full data migration

  • Enhanced DDL execution framework
    Provide a scalable parallel DDL execution framework to improve the performance and stability of DDL operations.
  • Table-level load balancing
    PD determines data scheduling strategies based on the workload of each Region on every table.
  • Improve performance of handling system tables with large data volumes
    Enhance query performance and reduce query overhead for system tables with large data volumes.
Reliability and Availability
Enhance dependability
  • Limit memory consumption for backup tasks

  • Limit memory consumption for statistics collection

  • Manage massive SQL bindings
    Improve the user experience of SQL binding, making it convenient for users to create and manage a large number of execution plans to stabilize database performance.

  • Enhance resource group control over complex SQL
    Regularly assess the Request Unit (RU) consumption of complex SQL before completion of SQL execution to prevent excessively large impacts on the entire system during execution.

  • Automatically switch resource groups for runaway queries
    When a query is identified as a runaway query, users can choose to switch it to a specific resource group and set an upper limit on resource consumption.
  • Limit memory consumption of schema metadata
    Enhance the stability of large-scale clusters.

  • Distributed statistics collection
    Statistics collection supports parallel execution across multiple TiDB nodes to improve collection efficiency.

  • Multi-version statistics
    After statistics are updated, users can view the historical versions and choose to restore them to an earlier version.

  • Reliable data backup
    Reduce potential issues like insufficient memory during data backup and ensure the availability of backup data.

  • Common operators support spilling to disk
    Common operators such as HashAgg, Sort, TopN, HashJoin, WindowFunction, IndexJoin, and IndexHashJoin support spilling to disk, reducing the risk of out-of-memory (OOM).
  • Adaptive resource group
    Resource groups automatically adjust their Request Unit (RU) settings based on past execution patterns.

  • Enhanced memory protection
    TiDB actively monitors the memory usage of all components and prevents memory operations that might impact system stability.

  • Instance-level execution plan cache
    All sessions within the same TiDB instance can share the execution plan cache, improving memory utilization.
Database Operations and Observability
Enhance DB manageability and its ecosystem
  • Reliable query termination
    Running SQL statements can be immediately terminated, and the corresponding resources are released from TiDB and TiKV.

  • Permission control for switching resource groups
    Only users with specific permissions can switch their resource groups, thus preventing resource abuse.

  • Mapping tables or SQL with hot Regions

  • Logical data import mode with IMPORT INTO

  • Fine-grained customization of statistics collection
    Users can modify the statistics collection strategy for specific tables, such as healthiness and parallelism.

  • Workload Repository
    TiDB persists workload information in memory to permanent volume, including cumulative and real-time statistic data, which aids in troubleshooting and analysis.

  • Automatic index advisor
    TiDB automatically analyzes SQL statements that can be optimized and recommends creating or dropping indexes.

  • Support modifying column types for partitioned tables
    Users can modify the data type of columns in partitioned tables, regardless of whether a column is a partitioning key.

  • Conflict strategy for IMPORT INTO
    Users can set the conflict resolution strategy when importing data, such as exiting with an error, ignoring, or replacing in case of conflicts.

  • End-to-End monitoring
    Track the time consumption of individual SQL statements throughout their entire lifecycle, including consumption on TiDB, TiKV, PD, and TiFlash components.
  • Workload analysis
    Analyze historical workload data from the Workload Repository and provide optimization recommendations, such as SQL tuning and statistics collection.

  • Revisable primary key

  • Export data as SQL statements

Security
Enhance data safety and privacy
  • Google Cloud KMS
    Enhance the key management mechanism for static encryption based on Google Cloud KMS, making it generally available (GA).

  • Improved dynamic privilege
    Improve the dynamic privilege design and limit the implementation of Super privilege.

  • Marker-based log desensitization
    Support marking sensitive information in the cluster log. Then, you can determine whether to desensitize it according to the usage scenario.

  • FIPS
    Encryption scenarios comply with FIPS.
  • IAM authentication for AWS
    TiDB as AWS third-party ARN for AWS IAM access.

  • Kerberos
    Support Kerberos-based authentication.
  • MFA
    Support the multi-factor authentication mechanism.
  • Label-based access control
    Support data access control by configuring labels.

  • Enhanced client-side encryption
    Support client-side encryption of key fields to enhance data security.

  • Dynamic desensitization of business data
    Support desensitizing data based on different data application scenarios to ensure data security in important fields.

Roadmap - 图1

Note

These are non-exhaustive plans and are subject to change. Features might differ per service subscriptions.