TiDB 2.0 Release Notes

On April 27, 2018, TiDB 2.0 GA is released! Compared with TiDB 1.0, this release has great improvement in MySQL compatibility, SQL optimizer, executor, and stability.

TiDB

  • SQL Optimizer
    • Use more compact data structure to reduce the memory usage of statistics information
    • Speed up loading statistics information when starting a tidb-server process
    • Support updating statistics information dynamically [experimental]
    • Optimize the cost model to provide more accurate query cost evaluation
    • Use Count-Min Sketch to estimate the cost of point queries more accurately
    • Support analyzing more complex conditions to make full use of indexes
    • Support manually specifying the Join order using the STRAIGHT_JOIN syntax
    • Use the Stream Aggregation operator when the GROUP BY clause is empty to improve the performance
    • Support using indexes for the MAX/MIN function
    • Optimize the processing algorithms for correlated subqueries to support decorrelating more types of correlated subqueries and transform them to Left Outer Join
    • Extend IndexLookupJoin to be used in matching the index prefix
  • SQL Execution Engine
    • Refactor all operators using the Chunk architecture, improve the execution performance of analytical queries, and reduce memory usage. There is a significant improvement in the TPC-H benchmark result.
    • Support the Streaming Aggregation operators pushdown
    • Optimize the Insert Into Ignore statement to improve the performance by over 10 times
    • Optimize the Insert On Duplicate Key Update statement to improve the performance by over 10 times
    • Optimize Load Data to improve the performance by over 10 times
    • Push down more data types and functions to TiKV
    • Support computing the memory usage of physical operators, and specifying the processing behavior in the configuration file and system variables when the memory usage exceeds the threshold
    • Support limiting the memory usage by a single SQL statement to reduce the risk of OOM
    • Support using implicit RowID in CRUD operations
    • Improve the performance of point queries
  • Server
    • Support the Proxy Protocol
    • Add more monitoring metrics and refine the log
    • Support validating the configuration files
    • Support obtaining the information of TiDB parameters through HTTP API
    • Resolve Lock in the Batch mode to speed up garbage collection
    • Support multi-threaded garbage collection
    • Support TLS
  • Compatibility
    • Support more MySQL syntaxes
    • Support modifying the lower_case_table_names system variable in the configuration file to support the OGG data replication tool
    • Improve compatibility with the Navicat management tool
    • Support displaying the table creating time in Information_Schema
    • Fix the issue that the return types of some functions/expressions differ from MySQL
    • Improve compatibility with JDBC
    • Support more SQL Modes
  • DDL
    • Optimize the Add Index operation to greatly improve the execution speed in some scenarios
    • Attach a lower priority to the Add Index operation to reduce the impact on online business
    • Output more detailed status information of the DDL jobs in Admin Show DDL Jobs
    • Support querying the original statements of currently running DDL jobs using Admin Show DDL Job Queries JobID
    • Support recovering the index data using Admin Recover Index for disaster recovery
    • Support modifying Table Options using the Alter statement

PD

  • Support Region Merge, to merge empty Regions after deleting data [experimental]
  • Support Raft Learner [experimental]
  • Optimize the scheduler
    • Make the scheduler to adapt to different Region sizes
    • Improve the priority and speed of restoring data during TiKV outage
    • Speed up data transferring when removing a TiKV node
    • Optimize the scheduling policies to prevent the disks from becoming full when the space of TiKV nodes is insufficient
    • Improve the scheduling efficiency of the balance-leader scheduler
    • Reduce the scheduling overhead of the balance-region scheduler
    • Optimize the execution efficiency of the hot-region scheduler
  • Operations interface and configuration
    • Support TLS
    • Support prioritizing the PD leaders
    • Support configuring the scheduling policies based on labels
    • Support configuring stores with a specific label not to schedule the Raft leader
    • Support splitting Region manually to handle the hotspot in a single Region
    • Support scattering a specified Region to manually adjust Region distribution in some cases
    • Add check rules for configuration parameters and improve validity check of the configuration items
  • Debugging interface
    • Add the Drop Region debugging interface
    • Add the interfaces to enumerate the health status of each PD
  • Statistics
    • Add statistics about abnormal Regions
    • Add statistics about Region isolation level
    • Add scheduling related metrics
  • Performance
    • Keep the PD leader and the etcd leader together in the same node to improve write performance
    • Optimize the performance of Region heartbeat

TiKV

  • Features
    • Protect critical configuration from incorrect modification
    • Support Region Merge [experimental]
    • Add the Raw DeleteRange API
    • Add the GetMetric API
    • Add Raw Batch Put, Raw Batch Get, Raw Batch Delete and Raw Batch Scan
    • Add Column Family options for the RawKV API and support executing operation on a specific Column Family
    • Support Streaming and Streaming Aggregation in Coprocessor
    • Support configuring the request timeout of Coprocessor
    • Carry timestamps with Region heartbeats
    • Support modifying some RocksDB parameters online, such as block-cache-size
    • Support configuring the behavior of Coprocessor when it encounters some warnings or errors
    • Support starting in the importing data mode to reduce write amplification during the data importing process
    • Support manually splitting Region in halves
    • Improve the data recovery tool tikv-ctl
    • Return more statistics in Coprocessor to guide the behavior of TiDB
    • Support the ImportSST API to import SST files [experimental]
    • Add the TiKV Importer binary to integrate with TiDB Lightning to import data quickly [experimental]
  • Performance
    • Optimize read performance using ReadPool and increase the raw_get/get/batch_get by 30%
    • Improve metrics performance
    • Inform PD immediately once the Raft snapshot process is completed to speed up balancing
    • Solve performance jitter caused by RocksDB flushing
    • Optimize the space reclaiming mechanism after deleting data
    • Speed up garbage cleaning while starting the server
    • Reduce the I/O overhead during replica migration using DeleteFilesInRanges
  • Stability
    • Fix the issue that gRPC call does not get returned when the PD leader switches
    • Fix the issue that it is slow to offline nodes caused by snapshots
    • Limit the temporary space usage consumed by migrating replicas
    • Report the Regions that cannot elect a leader for a long time
    • Update the Region size information in time according to compaction events
    • Limit the size of scan lock to avoid request timeout
    • Limit the memory usage when receiving snapshots to avoid OOM
    • Increase the speed of CI test
    • Fix the OOM issue caused by too many snapshots
    • Configure keepalive of gRPC
    • Fix the OOM issue caused by an increase of the Region number

TiSpark

TiSpark uses a separate version number. The current TiSpark version is 1.0 GA. The components of TiSpark 1.0 provide distributed computing of TiDB data using Apache Spark.

  • Provide a gRPC communication framework to read data from TiKV
  • Provide encoding and decoding of TiKV component data and communication protocol
  • Provide calculation pushdown, which includes:
    • Aggregate pushdown
    • Predicate pushdown
    • TopN pushdown
    • Limit pushdown
  • Provide index related support
    • Transform predicate into Region key range or secondary index
    • Optimize Index Only queries - Adaptively downgrade index scan to table scan per Region
  • Provide cost-based optimization
    • Support statistics
    • Select index
    • Estimate broadcast table cost
  • Provide support for multiple Spark interfaces
    • Support Spark Shell
    • Support ThriftServer/JDBC
    • Support Spark-SQL interaction
    • Support PySpark Shell
    • Support SparkR