AnalyticDB for PostgreSQL - Available in Public Preview
Oct 08 2019
AnalyticDB for PostgreSQLContent
Target customers: developers of Internet, games, and government industries, large retail chains, financial and insurance industries, and examination and certification institutions. Features released: Alibaba Cloud has officially released AnalyticDB for PostgreSQL 6.0 based on the latest kernel of Greenplum 6.0. The concurrent transaction processing capability of AnalyticDB for PostgreSQL 6.0 has been improved to meet real-time data warehousing scenarios. AnalyticDB for PostgreSQL 6.0 also supports HTAP services by optimizing features such as transaction locks, which improves the transaction processing capability by 70 times. The kernel of AnalyticDB for PostgreSQL 6.0 has been upgraded from PostgreSQL 8.2 to PostgreSQL 9.4, which is better compatible with the PostgreSQL community. New features of AnalyticDB for PostgreSQL 6.0 include: JSONB type: AnalyticDB for PostgreSQL 6.0 supports the JSON storage format as well as JSONB for high-performance JSON data processing and more JSON functions. UUID type: AnalyticDB for PostgreSQL 6.0 supports the UUID data type. GIN index and SP-GiST index: support fuzzy matching with higher performance. Fine-grained permissions control: supports schema-level and column-level permissions control and authorization. Efficient VACUUM method: When you use the VACUUM statement to release spaces, locked pages are temporarily skipped and vacuumed later when polling. This method can alleviate the overall blocking status. DBLink: supports cross-database query and access. Recursive CTE: The recursive query feature of SQL statements is enabled to process data of hierarchical or tree structure, facilitating multi-level recursive query for these types of data. PL/SQL enhancement: supports the RETURN QUERY EXECUTE statement to execute SQL statements in real time. The anonymous blocks definition is supported. New OLAP features Replicated table: You can use the DISTRIBUTED REPLICATED clause to create replicated tables for dimension tables that are in the data warehouse. This greatly reduces data transmission and improves the query efficiency. Zstandard compression algorithm: The Zstandard compression algorithm performs three times better than the zlib compression algorithm. HTAP (OLAP + OLTP) capability improvement The global deadlock detection mechanism collects and analyzes lock information to detect and remove global deadlocks. It then updates and modifies HEAP tables with fine-grained row locks. Highly concurrent modification and DELETE statements are supported, which improves the concurrency and throughput of the entire system. The transaction locks are optimized to reduce lock competition at the start and end of transactions. AnalyticDB for PostgreSQL 6.0 can also process high-throughput transactions based on the high-performance OLAP analysis. In typical OLTP scenarios, TPC-C reaches 100,000 transactions per minute, and Sysbench can reach 150,000 transactions per second for SELECT statements, 50,000 transactions per second for INSERT statements, and 20,000 transactions per second for UPDATE statements.