2022

You can upgrade an instance to the latest minor engine version in the AnalyticDB for PostgreSQL console. For more information, see Update the minor engine version.

July 28, 2022 (V6.3.10.1)

Category	Feature	Description	Related documents
New	Diagnostics and optimization	The AnalyticDB for PostgreSQL diagnostics and optimization feature now includes three pages: Overview, Key Metrics, and Query Analysis. Overview: You can view the connection status, cluster status, and storage status of an instance. Key Metrics: You can monitor data bloat, data skew, and index usage. Query Analysis: You can view the history and status of SQL queries and analyze query execution using detailed execution plans.	Instance status overview Data bloat, data skew, and index statistics Query diagnostics
	Real-time materialized view	Real-time materialized views now support: OUTER JOIN. Nested real-time materialized views. View-level query rewrite.	Real-time materialized view
Optimized	Real-time materialized view	Improves the write performance of real-time materialized views.	Real-time materialized view
Fixed	Fixes an issue where data writes fail if the user of a real-time materialized view is different from the user of the base table.
	Fixes an issue where an error occurs during automatic query rewrite for a materialized view that involves a window function.

April 15, 2022 (V6.3.9.0)

Category	Feature	Description	Related documents
New	SQL plan management	AnalyticDB for PostgreSQL introduces the SQL plan management feature. This feature stores the execution plan for a specific SQL statement. For similar SQL statements, the system can apply the stored execution plan with necessary adjustments. You can also view and manually modify the stored execution plans.	SQL plan management
Optimized	Hint	The hint feature is enabled by default.	Use hints
	Shared input scan operator	Improves the performance of the shared input scan operator for highly concurrent and complex Common Table Expression (CTE) queries.	None

February 21, 2022 (V6.3.8.0)

Category	Feature	Description	Related documents
New	Query profiling statistics	AnalyticDB for PostgreSQL introduces query profiling statistics. After you enable this feature, the system automatically collects and records statistics during query execution. You can view the statistics in system views.	Query profiling statistics
	Dynamic join filter	AnalyticDB for PostgreSQL introduces the dynamic join filter feature. This feature improves the performance of hash joins by dynamically collecting join key information from the right table of a hash join to filter out non-matching data from the left table before computing the join. This reduces unnecessary disk I/O, network traffic, and CPU overhead.	Dynamic join filter
	Data synchronization	You can use the data synchronization feature to synchronize data from an ApsaraDB RDS for PostgreSQL instance to an AnalyticDB for PostgreSQL instance.	Data synchronization
	Realtime Compute for Apache Flink integration	A new version of the Flink AnalyticDB for PostgreSQL Connector is now available. You can use Apache Flink to write data to or read data from AnalyticDB for PostgreSQL.	Write data to AnalyticDB for PostgreSQL by using Realtime Compute for Apache Flink Use Apache Flink to perform real-time reads and writes of full data
Optimized	Error messages for oversized fields	If you try to insert or update data that exceeds a field's defined length, the system now returns a detailed error message that identifies the oversized field.	None
	Bitmap scan	The bitmap scan path now includes statistics for index scan tuples. When you execute an SQL statement that has an execution plan containing a bitmap scan, the system accurately counts the number of times an index is used and the number of rows scanned.	Index management
	Partitioned table	A JIT cache is now available for partitioned tables to prevent repeated compilation. This improves query performance by seconds when the number of partitions is greater than 100.	Partitioned tables
	Laser computing engine	The Laser computing engine supports the MergeAppend operator. If subqueries are ordered and the final result also needs to be sorted by the same key, the MergeAppend operator uses a heap sort algorithm to directly output the merged results of the ordered subqueries.	Use the Laser computing engine
	OSS foreign table	OSS foreign tables support reading and writing data in the Avro format.	Use an OSS foreign table to access OSS data
	Backup and restoration	Improves the performance of backup and restoration with parallel archiving, which accelerates incremental backups.	Backup and restoration
Fixed	Fixes an issue where redistribution is triggered because the distribution key information is lost after a UNION ALL or similar operation is performed on a table whose distribution key is of the VARCHAR or TEXT data type.
	Backward scanning of tuplestores is now supported to prevent Backward scanning of tuplestores are not supported at this time errors for some queries.
	Fixes an issue where DDL operations fail on secondary coordinator nodes during an HA switchover between the primary and standby coordinator nodes.

January 7, 2022 (V6.3.7.0)

Category	Feature	Description	Related documents
New	pg_hint_plan extension	The pg_hint_plan extension is now available in AnalyticDB for PostgreSQL. You can use hints in the form of comments to control and optimize execution plans. You can also register the optimized SQL templates and hints. For subsequent SQL statements that match a registered template, the system automatically applies the optimized execution plan to improve efficiency.	Use hints Use hints to control execution plans
	Query cache	AnalyticDB for PostgreSQL introduces a query cache to store query results, reducing execution time to milliseconds for repeated queries.	Query cache for V6.0
Optimized	LLVM JIT	AnalyticDB for PostgreSQL introduces the LLVM JIT feature from PostgreSQL 14.	None
	Backup and restoration	AnalyticDB for PostgreSQL improves backup performance with parallel archiving.	Back up data
Fixed	Fixes an issue where the ORDER BY clause in a subquery is ignored if the parent query contains only a LIMIT clause.
	Fixes an issue that caused a rebalance operation to fail during cross-machine recovery of a compute node due to a missing mirror host entry in the pg_hba.conf file of the primary compute node.
	Fixes an issue where creating a partitioned table on the primary coordinator node fails in a multi-coordinator architecture.