Introduction: Alibaba cloud RDS PostgreSQL is a fully managed cloud database product compatible with open source PostgreSQL. Since its first release in 2015, it has been continuously upgraded and iterated according to user needs. It supports multiple versions such as 9.4, 10, 11 and 12, covering multiple forms such as highly available version, basic version and read-only instance. Data storage covers local SSD, SSD cloud disk ESSD cloud disk and other forms. The maximum configuration can reach 104 cores, 768g memory, 32tb and 1 million IOPs.
Alibaba cloud RDS PostgreSQL is a fully managed cloud database product compatible with open source PostgreSQL. Since its first release in 2015, it has been continuously upgraded and iterated according to user needs. It supports multiple versions such as 9.4, 10, 11 and 12, covering multiple forms such as highly available version, basic version and read-only instance. Data storage covers multiple forms such as local SSD, SSD cloud disk and ESSD cloud disk. The maximum configuration can reach 104 cores, 768g memory, 32tb and 1 million IOPs.
In order to meet the database demands of different industries, Alibaba cloud PostgreSQL has launched the hot and cold separation (OSS)_ FDW) stores external tables, ganos geographical space-time expansion module, Chinese word segmentation module, chemical analysis module, image recognition module, etc., which has served various industries such as e-commerce, enterprise ERP, banking, hotel, high technology, logistics, Internet, travel, multimedia, etc. At the beginning of 2021, we brought the latest heavyweight product upgrade release:
- RDS PostgreSQL 13 heavy release:
Compatible with the latest version 13 of the community, BTREE index is more efficient and space-saving, the performance of partitioned tables and aggregate queries is greatly improved, the function of JSON is more powerful, the extended statistical information is enhanced, better SQL optimization logic is supported, and incremental sorting and parallel index garbage collection functions are added.
- One click large version upgrade function release:
In order to enable users of the lower version to upgrade to the higher version more efficiently and enjoy more functions and high performance brought by the higher version, we have launched the one click large version upgrade function. Users of the lower version can upgrade to the larger version almost smoothly (the cutover process is short and read-only).
Many users choose the basic version based on early cost considerations or business importance. After the business importance is improved, they want to upgrade to the highly available version. Previously, it can only be realized by migrating data, which takes a long time, and the business connection string needs to be changed. The experience is not very friendly. Now we support one click upgrade to the highly available version.
- Solve the problem of DDL incremental subscription:
In terms of stability, Alibaba cloud RDS PostgreSQL solves the problem of logical incremental data synchronization: DDL cannot be captured and cannot be synchronized to the downstream. Therefore, as long as the main database executes DDL statements, when these statements involve structural changes or new structures, the downstream playback of incremental data generated after these DDLS may fail. Now we support DDL subscription. Combined with DTS service, we can support RDS PostgreSQL incremental data synchronization.
- Solve the problem of slot site loss of logical subscription in streaming replication ha environment:
There is a problem with the community version: the slot location information of logical incremental replication cannot be synchronized to the stream replication slave database. This problem will lead to the loss of slot information after the master-slave switch of the instance, which is also a headache for many PG database users. At the RDS PostgreSQL kernel level, we support the master-slave synchronization of the slot location. After the master-slave switch, the slot location information will not be lost and the logical replication is normal.
- Make RDS PostgreSQL more secure:
In terms of security, in this release, we launched the function of SQL firewall to support the learning mode, automatically learn the SQL executed in the database and record it in the white list. After the alarm is turned on, only the SQL in the white list can be executed, which can effectively prevent database deletion and running.
- Support es index:
In terms of flexibility, RDS PostgreSQL can better integrate other products, such as OSS_ FDW, users can access less accessed data on OSS as an external table, which is especially suitable for the storage of historical data and can help users save costs. This time, we will add integration with elasticsearch in RDS PostgreSQL. Why should we support this function? In fact, ES is a very popular search engine. Many users are using it, but users need to synchronize the data of relational database to es, which takes a lot of trouble. It can be realized through logical incremental replication, or the business layer can realize double writing by itself, which can not guarantee the requirements of consistency and delay. There are many problems. Now, users can create es indexes in RDS PostgreSQL tables. The data is stored in PostgreSQL. The indexes are in the ES cluster. Users do not need to maintain synchronization and consistency. They can write and query RDS PG normally. They will transparently use the capabilities of ES search engine to speed up the data search of RDS PG.
In terms of products, we continue to pursue the concept of faster, more stable and more secure, explore the real needs of customers, and provide customers with more satisfactory RDS PostgreSQL services.
Author: de Ge
This article is the original content of Alibaba cloud and cannot be reproduced without permission