At PGConf.Asia-Chinese Technology Forum, listen to the in-depth understanding of database technology from Tencent Cloud experts

In this PGConf.Asia 2021 Asia Technology Conference, in addition to the Tencent Cloud Database Sub-forum (focusing on PGConf.Asia-Tencent Cloud Sub-forum, Get financial-grade distributed database collection masterpiece), the Tencent Cloud database expert team will also be in the Chinese Technology Forum -The kernel session and the application practice session will bring 3 sharing sessions.

Live address:

https://2021.postgresconf.cn/liveEntry

Let's first take a look at the Tencent Cloud database product mentioned in the above speech.

TDSQL-C products:

Cloud Native Database TDSQL-C (Cloud NativeDatabase TDSQL-C) is referred to as TDSQL-C. TDSQL-C is a new generation of high-performance and highly available enterprise-level cloud native cloud database developed by Tencent Cloud. Combining the advantages of traditional databases, cloud computing and new hardware technologies, it is 100% compatible with MySQL and PostgreSQL, achieving a high throughput of over one million QPS, and 448TB massive distributed intelligent storage to ensure data security and reliability. The product has the following characteristics:

Fully compatible: 100% compatible with open source database engines MySQL and PostgreSQL. Almost no code changes are required to complete the smooth migration of existing database queries, applications and tools.

Ultra-high performance: It has the strong performance of commercial databases, the highest performance is 8 times that of MySQL database and 4 times that of PostgreSQL database.

Mass storage: A single instance supports 448TB of mass storage, which can easily cope with the continuous growth of business data.

Fast recovery: Compute nodes are stateless, support local and cross-device failover and recovery in seconds; support snapshot-based backup and rollback in seconds.

High data reliability: The cluster supports security group and VPC network isolation. Automatically maintain multiple copies of data and backups to ensure data safety and reliability, with a reliability of 99.9999999%.

Flexible expansion: Computing nodes can be quickly up and down according to business needs, and expansion can be completed in seconds. It supports serverless serverless architecture and automatically expands and shrinks to achieve the optimal cost of computing resources.

TDSQL-A products:

Analytical Datbase TDSQL-A (TDSQL-A) is a distributed analytical database system independently developed by Tencent, which provides you with efficient mass data storage and online analysis and processing capabilities. The product has the following characteristics:

Self-developed columnar storage: Provides efficient row and column hybrid query capabilities, provides columnar storage capabilities, has high compression ratios, and saves storage space.

High compatibility: support SQL 2003 and PostgreSQL syntax, highly compatible with Oracle syntax, support for stored procedures and complete distributed transaction capabilities.

Enterprise-level data security: It has enterprise-level security capabilities such as separation of powers, data desensitization, data encryption, and data auditing.

Mass data storage: MPP has no shared architecture, elastically expands nodes, and can support PB-level mass storage and large-scale cluster communication, large-scale cluster data interconnection, and related queries.

Efficient and complex query: Millisecond-level processing of trillion-wide table data, and second-level response of correlation analysis. Compared with open source and traditional databases, the performance is improved several times to hundreds of times; it has powerful OLAP analysis capabilities.

Powerful data management capabilities: Efficient and flexible online expansion, cold and hot separation of long-tail data, support for multiple types of partition tables, and support for secondary partition tables.

With the above knowledge to pave the way, let's look at the speeches of these three engineers separately:

The topic of Tencent Cloud expert engineer Sun Xu's speech is <<TDSQL-C PostgreSQL master-slave architecture optimization>>. Teacher Sun Xu has 10 years of database kernel development experience; familiar with database query processing, concurrency control, logging and storage systems; familiar with PostgreSQL (Greenplum, PGXC, etc.), Teradata and other database kernel implementation mechanisms. As we all know, the PostgreSQL database implements one-master and multiple-reading through data redundancy and log synchronization, and allows the standby machine to replace the host to continue to provide services to the outside world after the host is unavailable to ensure the availability of the system. However, TDSQL-C has a separate computing-storage architecture, and the database instances share the same data. Therefore, the design of one master and multiple reads is also very different from traditional databases. In this topic, Sun Xu will share with you the design and optimization of the master-slave architecture of TDSQL-C PostgreSQL. Database enthusiasts who are concerned about the design and optimization of the master-slave architecture of cloud native databases, please be sure to lock in the topic of this speech.

The topic of Tencent Cloud expert engineer Wu Xin's speech is "Technical Architecture Evolution of Enterprise Analytical Database TDSQL-A". Mr. Wu Xin has extensive experience in the fields of database, data replication, and big data computing, and has published many related papers and patents. Before joining Tencent, he worked in the IBM big data analysis team for many years, and then joined the Hashdata database startup company. After joining Tencent, he was responsible for the R&D-level design of multiple core functions such as TDSQL-A analytical database optimizer, executor, and distributed transaction. TDSQL-A is specifically designed for large-scale data scenarios in terms of kernel design and overall architecture, and it mainly solves the scalability problems of super-large-scale database clusters and the efficient analysis of large-scale data. This time, Mr. Wu Xin will share the detailed development history of enterprise-level analytical distributed databases, technical architecture innovation practices, and our future prospects for the future PG technology ecology.

The topic of the speech by Tencent Cloud Senior Engineer Huang Hui was "How to Build a Cross-Computer Room Disaster Recovery Practice for Ultra-Highly Available PostgreSQL". Teacher Huang Hui has been working in the database field since graduation, and is familiar with database high availability, management and control architecture, data backup and recovery, etc. Data reliability and database service availability are eternal topics in the category of database disaster recovery. Teacher Huang Hui’s sharing mainly focuses on how to balance between the two and some key trade-offs in high-availability design.

TIPS
At the main forum on December 14, Liu Shaorong, the head of the Tencent Cloud Database Technical Expert Group, will give you a speech on the evolution of Tencent Cloud PostgreSQL ecosystem and the direction of its technical architecture. Please look forward to revealing the secrets.

Special thanks