Data Management Expert: Transaction Processing

On June 26, 2019, GBASE (General Data Technology Co., Ltd., abbreviated as "GBASE") convened its 2019 Partner Conference under the theme "Phoenix Reborn, Sailing Anew" at the Millennium Hotel in Beijing. Over 100 customers and partners attended, including the State Information Center, General Administration of Customs, State Grid, Agricultural Bank of China (head office), Civil Aviation Information Center, Huawei, Digital China, Rongzhilian, H3C, and Weiwang Technology. Many attendees found the conference highly rewarding and wanted more insight. In response, today we present the second series: an in-depth introduction to transactional database products.
 

It is a well-known fact in the IT sector that transactional databases are the brightest jewel in the crown of data management and one of the most critical foundations of modern computing. They have thus become a strategic technological battleground for nations and enterprises. At the conference, GBASE unveiled two heavyweight transactional databases and one directory database, giving Chinese users options and solutions comparable to world-leading products.

Transactional Relational Database Cluster

Transactional databases have long been the weakest link in China's homegrown database sector. The transactional relational database cluster launched by GBASE will help highly complex environments, such as banks, reduce reliance on mainframes by offloading workloads and operations.

Transactional Relational Database GBase 8s MPP Cluster Architecture
 

Built on a highly reliable, mature transactional engine and combined with the company's years of expertise in MPP cluster technology, the transactional cluster leverages a federated architecture and elastic computing to deliver an enterprise-grade transactional relational database cluster. Compatible with domestic platforms, it compensates for performance limitations of domestic chips, achieving linear scalability as the cluster expands.

Trusted Secure Database

Trusted Secure Database GBase 8s V9 is a key outcome of the joint academy workstation established by GBASE and Academician Shen Changxiang. Adopting a dual architecture of traditional and trusted systems, it provides users with a fully trusted data computing environment. Compared with traditional security databases, the Trusted Computing 3.0-based secure database integrates trusted hardware, trusted operating systems, trusted networks, and other trusted environments, providing active immunity to effectively defend against attacks exploiting platform vulnerabilities, bugs, and SQL viruses.

Trusted Secure Database Architecture GBase 8s V9

Fully Trusted Data Computing Environment

Including a trusted access environment covering hardware, operating system, and runtime; a trusted network encompassing trusted client hardware and software; and clients with trusted drivers. Trusted static measurement covers programs, data, and configurations. Trusted dynamic measurement applies to SQL, functions, stored procedures, triggers, and other execution objects.

Trusted Computing Environment

Extraction and Measurement of Trusted Features

Feature Extraction: For each user, a set of legitimate rules for SQL statements, stored procedures, triggers, UDRs, etc. is stored. For each rule defined in the set, feature extraction and hash calculation produce a corresponding feature code, which is stored in the user-associated feature code library.

SQL Measurement: When a user requests statement execution, the statement is parsed to generate a syntax tree. The syntax tree undergoes rule analysis and feature collection, which is then compared with stored features; any mismatch leads to rejection.

Extraction and Measurement of Trusted Features

GBase 8s V9 integrates traditional security technologies required by China's Information Security Classified Protection Level 4 (MLPS 4) and next-generation security technologies such as the active immunity Trusted Computing 3.0 system, providing an optimal solution for users with stringent data security requirements.

High-Speed In-Memory Directory Database Cluster

As industries in China, especially public service and critical infrastructure, raise the bar for security checks, identity verification, and anti-counterfeiting, many large-scale users demand higher speed, greater concurrency, and support for a larger number of entries from traditional LDAP-based authentication and signature verification systems. GBase XDM Cluster, the high-speed in-memory directory database cluster, represents a major upgrade of GBASE's traditional strength in LDAP technology, addressing these urgent requirements. Through clustering, in-memory computing, high availability, read/write splitting, and load balancing, GBase XDM Cluster meets the needs of large-scale PKI/CA infrastructures, large multimodal authentication platforms, large 4A (Authentication, Authorization, Accounting, Audit) systems, and large-scale platforms requiring ultra-fast, precise queries.

GBase XDM Cluster – Product Architecture

GBase XDM Cluster components back up one another and achieve load balancing via APIs. Logically, the cluster presents as a single entity, with data distributed across nodes, and multiple logical entities can be deployed for mutual backup. Vertically, compute nodes are backed up, and cluster components schedule compute nodes to achieve high availability and load balancing. Flexible configuration is supported, and the cluster can be scaled online by adding compute nodes for hot data shards while in production.

High Availability/Load Balancing

The strongest advantage of GBase XDM Cluster is its extreme performance. In a configuration with 4 Intel Xeon E7-4850V4 processors, 512 GB DDR4 memory, hot-swappable SAS drives (≥10,000 RPM, single disk capacity ≥1.2 TB), and 10 Gbps network bandwidth, it achieves up to 368,000 QPS per node, 1,794,000 QPS with 5 nodes, and 3,355,000 QPS with 10 nodes, demonstrating linear scalability as nodes are added.

Applicable scenarios:

Large-scale PKI/CA infrastructures

Large multimodal authentication platforms: certificate recognition, facial recognition, iris recognition, fingerprint recognition, voiceprint recognition, etc.

Large 4A authentication systems

Large-scale system platforms

Ultra-fast, precise queries