PolarDB:Release notes for PolarDB-X columnar nodes

Category

Description

Feature optimization

The performance of calculating routing partitions is optimized for IMCIs.

The columnar engine is optimized to improve stability.

Fixed issue

The following issue is fixed: The data synchronization between data nodes and columnar nodes stops when you delete non-IMCI indexes by executing the DROP INDEX statement.

The following issue is fixed: After the CDC client is interrupted and then reconnected, the consumer offset of binary logs is incorrect and the data does meet your expectations.

Category

Description

New feature and feature optimization

Versions are supported for primary keys and sort keys of existing IMCIs.

IMCI supports sorting data by the sort key in descending order.

The performance of calculating partition routing is optimized for IMCIs.

The process of deleting metadata of read-only column store instances is optimized to reduce the resource consumption of Global Meta Service (GMS) nodes.

The engine stability is optimized.

Fixed issue

The following issue is fixed: When you delete non-IMCI indexes by executing the DROP INDEX statement, the synchronization between the data node and columnar node stops.

The following issue is fixed: When data of the decimal type is expressed in the scientific notation format, the IMCI fails to extract full data.

The following issue is fixed: When the columnar storage engine is idle, the adaptive heartbeat interval becomes longer.

The following issue is fixed: IMCI creation fails because the target table contains a hidden column named TSO or POSITION.

Category

Description

Feature optimization

The process of deleting metadata of read-only column store instances is optimized to reduce the resource consumption of Global Meta Service (GMS) nodes.

Fixed issue

The following issue is fixed: When the columnar storage engine is idle, the adaptive heartbeat interval keeps growing.

Category

Description

New feature

IMCI supports sorting data by the sort key in descending order. For more information, see Create and use a CCI.

Fixed issue

The following issue is fixed: When the columnar storage engine is idle, the adaptive heartbeat interval becomes longer.

Category

Description

New feature and feature optimization

Tables that contain IMCIs support the TRUNCATE TABLE statement.

The lease time is optimized for various node specifications to avoid frequent HA switchovers caused by incorrect lease time configurations.

The compaction operation of column-oriented data is optimized to avoid generating a large amount of small files.

Category

Description

New feature

IMCI supports partition splitting. For more information, see Split a partition into multiple partitions.

Category

Description

Fixed issue

The following issue is fixed: IMCI data is incorrectly created if the primary key is of the BIGINT UNSIGNED type.

The following issue is fixed: The columnar node stops running when a table contains a large number of partitions and concurrent access requests are made to the OSS client.

Category

Description

Fixed issue

The following issue is fixed: In a read-only column store instance, certain primary key values cannot be retrieved after the corresponding key-value pairs are added to a HashMap.

Modified

Previously, DDL operations on primary tables with columnstore indexes were prohibited by default. Now, DDL operations are allowed by default on such tables.

Category

Description

New feature and feature optimization

The files on the columnar nodes that exceed the specified retention period can be automatically cleared.

The compression performance of CSV files is improved.

The initialization performance of columnar nodes is improved when highly concurrent data compression tasks are performed.

Fixed issue

The following issue is fixed: When you synchronize incremental data from the primary instance to a column store instance, the synchronization is interrupted by the RENAME statements.

Category

Description

New feature and feature optimization

Adaptive low latency, batch commits, cache control, and incremental, full, and parallel data pull are supported.

Persistence, table-level separation, and high-performance bulk operations are supported.

Index creation and deletion, simple column alterations, and simple data type conversions are supported.

Asynchronous scheduling, concurrency control, read/write parallelism, and more than 10 types of compaction methods are supported.

Fast recovery upon restart, dual-node disaster recovery, and utilization of resources on secondary nodes are supported.

Object Storage Service (OSS), Network File System (NFS), and custom cache systems can be used for storage.

All MySQL data types are supported for new data columns. Common data types are supported for primary keys, sort keys, and partition keys.

Simple verification of row and column data consistency is supported.

Status monitoring, column store data monitoring, and exception alerting are supported.