The release of the distributed database Apache Cloudberry 2.0.0 has been announced, continuing the development of the open source codebase of the Greenplum database, which was turned into a proprietary product by Broadcom after acquiring VMware. Apache Cloudberry 2.0.0 is marked as the first release of the project following the transfer of the code to the Apache community. The project is currently in the Apache incubator and will be transitioned to a primary Apache project once the infrastructure and support are ready.
The Cloudberry DBMS is a distributed edition of the open-source PostgreSQL DBMS, optimized for executing analytical queries over large datasets (Data Warehouse). It utilizes a massively parallel processing (MPP) architecture for parallel data processing, ensuring the scalability of the repository to petabyte sizes by segmenting data and employing a cluster of nodes for their storage and processing. servers.
Among the improvements in Apache Cloudberry 2.0.0:
- Migration to the PostgreSQL 14 codebase has been completed (Greenplum was based on PostgreSQL 12).
- Support for dynamic tables has been added, which allow for automatic updates of query results. This new feature is useful when working with real-time data, such as for real-time data analysis, Lakehouse architectures, and automated ETL (Extract, Transform, Load) handlers.
- A new hybrid storage format called PAX (Partition Attributes Across) has been implemented, combining the capabilities of row-based and column-based storage. The PAX format allows for high performance under both high write loads and during analytical query execution.
- The planning and execution of distributed queries have been significantly optimized.
- Resource management has improved: memory and CPU efficiency on cluster nodes has increased.
- Data distribution across nodes and parallel query processing have been enhanced.
- Backup strategies in distributed environments have been expanded.
- Licensing information and the formatting of code file headers have been aligned with Apache Foundation requirements. Binary artifacts have been excluded from the releases in the source code.
- The process of building project components in C++ and Python has been improved.
Source: opennet.ru
