Fun Talk: GBase 8a Database Cluster (I)

Hello to all GBase 8a fans and anyone interested in it.

I plan to take a deep dive into GBase 8a database cluster software—its development history, product architecture evolution, feature evolution, applicable scenarios, and comparative analysis with competitors—through a series of "Fun Talks." I hope you'll follow along and gain deeper insights into GBase 8a from these articles.

First, a disclaimer: this series is my original work. If you find any inaccuracies, omissions, or errors in the descriptions, your feedback and corrections are most welcome.

Without further ado, let's kick off the first "Fun Talk" by exploring what GBase 8a is, where it comes from, and where it is heading.

(I) What is GBase 8a?

The full name of GBase 8a is GBase 8a MPP Cluster. It is a core independently developed product of General Data Technology Co., Ltd. (GBASE). GBase 8a is an analytical database product, i.e., an OLAP database, primarily applied in data warehouses, data marts, business intelligence, and decision support systems. It has passed China's national evaluation for independently controllable technology and is fully verified as such. As a self-developed product, from the start of its R&D to the present, it has continuously collected and adopted customer requirements, added new features, and refined existing functionalities. After more than a decade of development and multiple major version iterations, it has achieved large-scale deployment in the finance and telecom industries. GBase 8a is a mature, fully domestically developed, high-end OLAP database product that Chinese users can trust.

(II) Where does GBase 8a come from?

The evolution of GBase 8a MPP Cluster can be divided into three phases: the columnar database phase, the MPP + columnar database phase, and the Logical Data Warehouse (LDW) phase. Each phase laid the R&D foundation for the next.

(1) Columnar database phase: This stage saw the birth of GBase 8a's core capabilities, including columnar storage, data compression, intelligent indexes, and parallel execution. General Data Technology started developing a columnar analytical database for big data in 2008, initiating the basic version of GBase 8a MPP Cluster, and officially released the standalone version of GBase 8a in June 2009.

(2) MPP + columnar database phase: This period marked the growth, expansion, and maturation of GBase 8a MPP Cluster through three major version iterations. In version V8.3, based on the standalone edition, a large-scale distributed cluster database was built, achieving distributed SQL, high-speed distributed loading, high availability, load balancing, online expansion, and backup/restore capabilities. The first version of GBase 8a MPP Cluster was officially released in May 2011. Subsequent evolution delivered two major iterations: V8.5 and V8.6. In April 2014, GBase 8a MPP Cluster V8.5 was released with significantly enhanced functionality and performance, including dramatically improved parallel processing, strong data consistency, support for up to 100 nodes, heterogeneous data integration, and built-in full-text search. It was adopted and rigorously tested by key customers such as Agricultural Bank of China. In 2015, version V8.6 was developed to support clusters of over 300 nodes, upgrading the architecture to a federation model that separates management clusters from data clusters. It also introduced multi-shard deployment, dynamic expansion, and online node replacement, further raising the node scale limit. In 2016, the product added access to and unified management of heterogeneous databases. By then, GBase 8a MPP Cluster had become a mature domestic OLAP database serving customers in telecom, finance, government, and enterprise sectors.

(3) Logical Data Warehouse (LDW) phase: This phase saw GBase 8a MPP Cluster comprehensively surpassing comparable foreign products. In 2018, GBase 8a v9, a large-scale virtual cluster version, was developed, supporting over 1,000 nodes, virtual clusters, and partitioned tables. In 2019, deep optimization for domestic servers began with version V9.5, which was extensively adapted and tuned for the domestic ecosystem—supporting Chinese chips like Kunpeng, Hygon, Phytium, Sunway, Loongson, and Zhaoxin, as well as domestic operating systems like Kylin, Founder, Deepin, UOS, and Ningsi. It supports multi-instance deployment for heterogeneous hardware active-active configurations; on all-domestic platforms, multi-instance deployment can more than double overall performance. It also supports active-active disaster recovery across Intel and domestic chip platforms. Moreover, GBase 8a V9.5 made substantial improvements in performance, functionality, and security. The CBO optimizer was enhanced, and batch loading for small files, high-concurrency queries, and OLAP analytical functions all saw multi-fold performance gains. Ultimately, in December 2020, it topped the TPC-DS benchmark with the best per-core performance.

(III) Where is GBase 8a heading?

GBASE focuses on database software products and services, striving to become the most trusted database product provider for its customers. As GBASE's core self-developed product, GBase 8a continues to evolve alongside big data technologies. It has grown, matured, and surpassed comparable foreign products such as Vertica and Greenplum, steadily gaining recognition and affirmation from customers in China, and officially stepping into the ranks of the most trustworthy domestic OLAP databases. Going forward, GBase 8a will closely follow the trends of both domestic and international big data technologies, continuously innovating and revolutionizing, to become the most advanced, stable, and performant analytical database with independent intellectual property rights for customers in China.

Wasn’t the title “Fun Talk”? Where is the fun, you ask? For those interested in GBase products, I’d say the fun is already there while you're reading this. So, let's talk—what would you like the next article to cover? It takes both "fun" and "talk" to make it a genuine "Fun Talk."