This topic describes the engine release notes for AnalyticDB for PostgreSQL in elastic storage mode in 2023 and provides links to the relevant references.
You can update your instances to the latest minor engine version in the AnalyticDB for PostgreSQL console. For more information about how to update the minor engine version, see Update the minor engine version.
December 2023 (V7.0.4.1)
Category | Description |
Fixed issue | The issue that Beam Client SDK does not support the COPY ON CONFLICT statement is fixed. |
The issue that causes core dumps when Query Monitor does not work is fixed. | |
The memory issue that occurs during index scanning is fixed for the Beam storage engine. |
December 2023 (V7.0.4.0)
Category | Feature | Description | References |
New feature | SM4 | The SM4 encryption algorithm can be used to encrypt and decrypt critical fields. | |
Delta support for dictionary encoding | Data in the Delta format can be optimized by using dictionary encoding to improve storage and query performance. | ||
Interleaved sorting | The Beam sorting optimization feature is supported. You can use compound sort keys and interleaved sort keys to implement more flexible and precise sorting capabilities. An interleaved sort key assigns an equal weight to each column in the sort key and is suitable for queries whose filter conditions contain a sort key column. | ||
btree-gist extension | The btree-gist extension provides a Generalized Search Tree (GiST) index structure to simulate B-tree behavior. Common B-tree search operators and the NotEqual ( | None | |
Warehouse intelligence | A language inference function is supported. You can use the function to interact with the large language models (LLMs) that are deployed in Elastic Algorithm Service (EAS) of Platform for AI (PAI) to implement warehouse intelligence. | None | |
Fixed issue | The Orca CLeftJoinPruning Rule issue is fixed. | ||
The issue that causes disordered tuple identifiers (TIDs) when you use Beam to create Bitmap indexes is fixed. | |||
The issue that SQL statements are truncated in Simple Log Service logs is fixed. | |||
The issues related to Laser Motion, Bitmap Index Scan, and Agg operators are fixed. | |||
The improper configuration issue of the work_mem parameter is fixed. | |||
November 2023 (V7.0.3.0)
Category | Feature | Description | References |
New feature | Tiered storage | The tiered storage of hot and cold data is supported. You can store data tables in Object Storage Service (OSS) to reduce storage costs. |
October 2023 (V7.0.2.4)
Category | Feature | Description | References |
New feature | pg_stat_statements and btree-gist extensions | The pg_stat_statements and btree-gist extensions can be used to analyze query execution records and handle searches that involve complex data types. | None |
Fixed issue | The issue that memory errors are caused by Laser Scan and Agg operators is fixed. | ||
The issue that Beam is deadlocked due to relation cache failures during Data Transmission Service (DTS) data synchronization is fixed. | |||
The issue that causes incorrect query results due to missing filter conditions of a subquery in the query optimizer is fixed. | |||
The issue that causes Agg operator errors due to unsegmented Hash Join Const results is fixed. | |||
The issue that causes unreasonable plans to be generated by the Orca optimizer is fixed. | |||
The issue that causes errors in reading Beam pos files and toast data is fixed. | |||
September 2023 (V7.0.2.0)
Category | Feature | Description | References |
New feature | Beam | Beam is a next-generation storage engine developed in-house by Alibaba Cloud based on table access methods of PostgreSQL 12 for AnalyticDB for PostgreSQL. Beam is designed to handle online transaction processing (OLTP) and online analytical processing (OLAP) workloads. OLTP involves high-concurrency reads and writes. OLAP involves batch writes and large-scale scanning. | |
Laser | Laser is a compute engine developed in-house by Alibaba Cloud for AnalyticDB for PostgreSQL and helps improve the performance of complex computing. In AnalyticDB for PostgreSQL V7.0, the Laser compute engine can accelerate the computing of Scan, Motion, Agg, Sort, and NestLoopJoin operators. | ||
Fixed issue | The inappropriate behavior of the ANALYZE statement for foreign tables is fixed. | ||
The exception issue of the gpexpand utility is fixed. | |||
The issue that causes inappropriate results of hash joins on CHAR-type columns is fixed. | |||
Multiple bugs of the pg_dump command are fixed. | |||
The memory leak issue of the auto_explain extension is fixed. | |||
The memory exception issue when the ADD Column statement is executed for append-optimized column-oriented (AOCO) tables is fixed. | |||
The gppkg package that has security risks is deleted. | |||
May 2023 (V7.0.1.8)
Category | Feature | Description | References |
New feature | Dynamic data masking | The dynamic data masking feature is supported. After this feature is enabled, sensitive data is masked in query results. | |
Optimized feature | adbpg toolkit extension | The adbpg toolkit extension can be used to optimize performance diagnostics. | None |
Fixed issue | The CVE-2023-2454 vulnerability is fixed. We recommend that you update the minor version of your AnalyticDB for PostgreSQL V7.0 instance in elastic storage mode to V7.0.1.8 or later. | ||
The issue that causes errors when you use the gptransfer utility is fixed. | |||
Inappropriate settings of the active_statements parameter in the pg_resqueue system table are corrected. | |||
The issue that causes disk storage errors of aggregation operations in scenarios that involve multiple grouping sets is fixed. | |||
The issue that causes the rds_superuser account to fail to create, modify, and delete resource groups is fixed. | |||
The issue that causes instance scale-down errors is fixed. | |||
The IN LIST filter condition can be converted into a join operation with temporary tables in the Orca optimizer. By default, this feature is disabled. You can enable this feature by using the | |||
The cost parameters of the left and right tables for hash joins are optimized. This prevents the Orca optimizer from producing execution plans that use large tables as right tables. | |||
The cost penalty threshold of broadcasting is modified in the Orca optimizer. This way, the optimizer can determine whether to use an execution plan that contains broadcast motions in a more accurate manner. | |||
January 2023 (V7.0.1.2)
Category | Feature | Description | References |
New feature | Change of compute node configurations | Database data can be read when you change the number of compute nodes. | |
The number of compute nodes can be reduced. | |||
PostGIS extension | The PostGIS extension is supported. | ||
Auto-vacuum | The auto-vacuum feature is supported to automatically execute VACUUM statements. | ||
Optimized feature | Orca optimizer | The Orca optimizer is optimized for distinct qualified aggregates (DQAs).
| None |
Transaction management | The default value of the idle_in_transaction_session_timeout parameter is 12. Unit: hours. Long-running transactions that remain inactive for more than 12 hours are automatically released. This prevents VACUUM operations from being blocked by idle transactions. | None | |
Fixed issue | The issue that causes scaling interruptions due to incorrect passwords is fixed. | ||
The issue that causes display failures of SQL audit data due to logging bugs is fixed. | |||
The issue that causes aggregation failures due to inappropriate execution plans generated by the Orca optimizer is fixed. | |||
The log level is decreased for the auto-vacuum feature to reduce the amount of logs. | |||