2023

Category

Description

Fixed issue

Beam Client SDK: fixed missing support for the COPY ON CONFLICT statement.

Fixed issue

Query Monitor: fixed core dumps that occurred when Query Monitor was inactive.

Fixed issue

Beam storage engine: fixed a memory issue during index scanning.

Category

Feature

Description

References

New feature

SM4

Encrypt and decrypt critical fields using the SM4 encryption algorithm.

Use pgcrypto and SM4

New feature

Delta support for dictionary encoding

Apply dictionary encoding to Delta-format data to improve both storage efficiency and query performance.

Dictionary encoding (public preview)

New feature

Interleaved sorting

Use compound sort keys and interleaved sort keys for more precise query filtering. An interleaved sort key assigns equal weight to each column, making it effective when filter conditions target any column in the sort key.

Beam sorting optimization (V7.0)

New feature

btree-gist extension

Use a Generalized Search Tree (GiST) index structure that simulates B-tree behavior. Supports common B-tree search operators and the NotEqual (<>) operator.

btree_gist

New feature

Warehouse intelligence

Interact with large language models (LLMs) deployed in Elastic Algorithm Service (EAS) of Platform for AI (PAI) through a language inference function.

Warehouse Intelligence (Public Preview)

Fixed issue

Fixed the Orca CLeftJoinPruning Rule issue.

Fixed issue

Fixed disordered tuple identifiers (TIDs) when using Beam to create bitmap indexes.

Fixed issue

Fixed truncated SQL statements in Simple Log Service logs.

Fixed issue

Fixed issues with Laser Motion, Bitmap Index Scan, and Agg operators.

Fixed issue

Fixed improper configuration of the work_mem parameter.

Category

Feature

Description

References

New feature

Tiered storage

Store data tables in Object Storage Service (OSS) to separate hot and cold data, reducing overall storage costs.

Tiered storage of hot and cold data

Category

Feature

Description

References

New feature

pg_stat_statements and btree-gist extensions

Analyze query execution records with pg_stat_statements, and handle searches involving complex data types with btree-gist.

—

Fixed issue

Fixed memory errors caused by Laser Scan and Agg operators.

Fixed issue

Fixed a Beam deadlock caused by relation cache failures during Data Transmission Service (DTS) data synchronization.

Fixed issue

Fixed incorrect query results caused by missing subquery filter conditions in the query optimizer.

Fixed issue

Fixed Agg operator errors caused by unsegmented Hash Join Const results.

Fixed issue

Fixed unreasonable execution plans generated by the Orca optimizer.

Fixed issue

Fixed errors when reading Beam pos files and toast data.

Category

Feature

Description

References

New feature

Beam

Handle both online transaction processing (OLTP) and online analytical processing (OLAP) workloads with Beam, a next-generation storage engine developed in-house by Alibaba Cloud. Beam is built on PostgreSQL 12 table access methods and optimized for high-concurrency reads and writes as well as batch writes and large-scale scanning.

Beam overview

New feature

Laser

Accelerate complex query computing with Laser, a compute engine developed in-house by Alibaba Cloud. In V7.0, Laser accelerates Scan, Motion, Agg, Sort, and NestLoopJoin operators.

Use the Laser computing engine

Fixed issue

Fixed inappropriate behavior of the ANALYZE statement for foreign tables.

Fixed issue

Fixed an exception in the gpexpand utility.

Fixed issue

Fixed incorrect results for hash joins on CHAR-type columns.

Fixed issue

Fixed multiple bugs in the pg_dump command.

Fixed issue

Fixed a memory leak in the auto_explain extension.

Fixed issue

Fixed a memory exception when executing ADD COLUMN for append-optimized column-oriented (AOCO) tables.

Fixed issue

Removed a gppkg package that posed security risks.

Category

Feature

Description

References

New feature

Dynamic data masking

Mask sensitive data in query results automatically after enabling this feature.

Dynamic data masking

Optimized feature

adbpg toolkit extension

Use the adbpg toolkit extension for more effective performance diagnostics.

—

Fixed issue

Fixed the CVE-2023-2454 vulnerability. 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.

Fixed issue

Fixed errors when using the gptransfer utility.

Fixed issue

Corrected inappropriate settings of the active_statements parameter in the pg_resqueue system table.

Fixed issue

Fixed disk storage errors in aggregation operations involving multiple grouping sets.

Fixed issue

Fixed an issue that prevented the rds_superuser account from creating, modifying, or deleting resource groups.

Fixed issue

Fixed instance scale-down errors.

Fixed issue

The Orca optimizer can now convert IN LIST filter conditions into a join with temporary tables. This is disabled by default; enable it with the adbpg_optimizer_enable_transform_in_list_to_semi_join parameter.

Fixed issue

Optimized hash join cost parameters so the Orca optimizer no longer uses large tables as the right-hand side of a join.

Fixed issue

Adjusted the broadcast motion cost penalty threshold in the Orca optimizer for more accurate execution plan selection.

Category

Feature

Description

References

New feature

Change of compute node configurations

Read database data while changing the number of compute nodes, including scale-down operations.

Change compute node configurations

New feature

PostGIS extension

Analyze geospatial data using the PostGIS extension.

Use PostGIS

New feature

Auto-vacuum

Automatically run VACUUM statements with the Auto-vacuum feature.

Configure scheduled maintenance tasks to clear junk data

Optimized feature

Orca optimizer

Improved performance for distinct qualified aggregates (DQAs): the optimizer now skips aggregate operators that cannot meet requirements, adds pre-aggregation stages during aggregation, and reduces the number of data redistributions.

—

Optimized feature

Transaction management

The idle_in_transaction_session_timeout parameter now defaults to 12 hours. Transactions inactive for longer than 12 hours are automatically released, preventing VACUUM operations from being blocked by idle transactions.

—

Fixed issue

Fixed scaling interruptions caused by incorrect passwords.

Fixed issue

Fixed display failures of SQL audit data caused by logging bugs.

Fixed issue

Fixed aggregation failures caused by inappropriate execution plans from the Orca optimizer.

Fixed issue

Reduced the log level for autovacuum to decrease overall log volume.