GBase 8a MPP Cluster V9 Feature Analysis
GBase 8a MPP Cluster, independently developed by GBase, is a mature analytical MPP database for the big data era. The latest GBase 8a MPP Cluster V9 virtual cluster features federation architecture, massively parallel processing, massive data compression, efficient storage structures, intelligent indexing, virtual clusters and mirroring, flexible data distribution, comprehensive resource management, online scaling, online node replacement, high concurrency, high availability, high security, easy maintenance, and high-speed loading. It meets the growing demands for data analysis, data mining, data backup, and ad-hoc queries in data-intensive industries.
New Features of GBase 8a MPP Cluster V9 Virtual Cluster
Enhanced High Availability
GBase 8a MPP Cluster provides multi-level high availability technologies, including process-level HA: the GCMonit component monitors service processes in real time to ensure healthy operation; and node-level HA to ensure data consistency.
Building on the existing multi-level HA, the new version uses virtual cluster mirroring to achieve real-time data consistency between mirrored clusters, enabling an active-active architecture and enhanced high availability at the database level. This provides an optional solution for cluster disaster recovery and read-write separation between data warehouse and data marts.
Virtual Clusters
The virtual cluster feature of GBase 8a MPP Cluster greatly enhances manageability, enabling a single cluster to scale up to thousands of nodes. Built on large-scale cluster architecture, a cluster can contain multiple Virtual Clusters (VCs). Each VC runs independently within the overall cluster, sharing a unified entry point. With proper permissions, VCs can access each other.
Unified Management
By partitioning a cluster into virtual clusters, scalability is enhanced while a unified management view is provided, allowing the cluster to reach thousand-node scale.
Unified Entry Point
Users can choose any coordinator node in the coordinator cluster as the entry point. The coordinator node determines the default VC based on the user.
Workload Isolation
The virtual cluster feature of GBase 8a MPP Cluster enforces vertical resource isolation, enabling physical isolation of different workloads in real-world business scenarios;
Transparent Access
It provides a unified access point (enabling access to both data warehouse and data marts), making access to multiple physical clusters transparent to applications—they appear as a single unified cluster.
Elevated Data Security
GBase 8a MPP Cluster provides data-at-rest encryption to meet security requirements and enhance system security. Encryption is performed at the Data Cell (DC) block level, supporting encryption at the table or column level.
Data Encryption
Dynamic Data Masking
Dynamic data masking does not alter the actual data stored in tables; it controls the data returned by queries. Its enforcement depends on user privileges: users with the unmask privilege can access actual data unaffected, while users without the unmask privilege see only masked data.
Dynamic Data Masking
Non-root User Installation and Operation
For ease of deployment, a one-click deployment script, SetSysEnv.py, is provided to handle operations requiring root privileges. Before cluster deployment, copy this script to all nodes (execute once for composite nodes) and run it under the root user on each node. Non-root installation and operation meet system-level security requirements.
Non-root User Installation and Operation
Other New Features and Improvements in GBase 8a MPP Cluster V9
Multi-Column Hash Distribution
In real-world scenarios where multi-column hash keys are used as primary keys, after migrating tables to GBase 8a, you can use these primary key columns as the hash columns. When joining on primary and foreign keys, the join can be performed on all columns locally on each node.
Multi-column hash distribution resolves data skew issues, preventing storage and computation skew on individual nodes that could create bottlenecks.
Support for GB18030 and UTF8MB4 Character Sets
The new version extends support for the four-byte encoding portion of GB18030 and UTF8MB4 encoding, addressing issues with rare, uncommon, and difficult characters, as well as meeting the pressing need across industries for characters used in personal and place names.
Performance Improvements
The new version includes optimizations in the following areas, leading to significant performance gains:
● Performance optimization for loading batches of small files
● Parallel optimization of OLAP functions
● Metadata merges
Partitioned Tables
Partitioning is an effective technique for handling large data volumes. It enhances availability, reduces maintenance costs, balances disk I/O, and improves query performance.
Tablespaces
GBase 8a tablespaces allow user-configurable data storage paths for Express engine tables; each tablespace corresponds to a storage path. Multiple tablespaces can exist per database, with one default tablespace. A tablespace can be used by multiple tables, but a table can belong to only one tablespace.
Auto-Increment Columns
An auto-increment column has the auto_increment attribute, with values increasing monotonically (not necessarily sequential). It can be used on columns of type tinyint, smallint, int, and bigint. An auto-increment column uniquely identifies each row, facilitating query, update, and delete operations. An auto-increment column uniquely identifies each row, facilitating query, update, and delete operations.