GBASE Financial Application Guide 1 | Overview of Distributed Analytical Databases
For two decades, GBASE has focused on database R&D, and its flagship GBase database has undergone 20 years of continuous iteration, forming a mature portfolio of database products for diverse business scenarios. Notably, GBase 8a MPP Cluster, released in 2011, as a large-scale distributed database cluster, has been continuously upgraded and refined over the past decade, staying aligned with international mainstream standards. It has been widely deployed in finance, telecom, government, energy, transportation, and other sectors, delivering many core-system construction and modernization cases for high-end customers. It effectively meets the needs of data-intensive industries for massive high-value data analysis, mining, backup, and ad-hoc querying.
To assist financial institutions in selecting distributed analytical database products and to share successful deployment experience in the financial industry, GBASE, under the guidance of the Beijing Fintech Industry Alliance, has compiled the 'GBase 8a Financial Application Guide'. This guide provides an in-depth look at the entire deployment process of a distributed analytical database—from selection planning and development design standards to data security, high availability, and operational optimization. It also introduces representative deployment cases of GBase 8a MPP Cluster at national policy banks and major state-owned banks.
GBASE will launch a series of articles to share insights from the Guide, hoping to provide reference for financial users in database selection and to support the efficient implementation and high-quality development of technology finance.
This article is the first in the series, covering the characteristics and classification of distributed analytical databases and their application scenarios in the financial industry.
Overview of Distributed Analytical Databases
1. Characteristics and Classification of Distributed Analytical Databases
With the development of the digital economy, data as a factor of production has become a fundamental economic resource alongside land, labor, capital, and technology. Data asset value grows with data volume, generating new demands for data storage and application across industries. Traditional data warehouse models and technologies face many challenges.
Challenge 1: Analytical systems handle large-scale data that changes rapidly and grows fast, with persistent storage needs. Data warehouses must store massive historical data and support online scaling.
Challenge 2: Business changes in analytical systems are increasingly frequent and trending toward intelligence, needing to support ad-hoc queries and data value discovery. Data warehouses must store comprehensive, fine-grained data and support data mining and machine learning.
Challenge 3: Given the uncertainty of analytical business, traditional Schema-on-Write modeling is time-consuming upfront and inflexible later. Data warehouses need to store all raw data and support late modeling.
Challenge 4: Storing all raw data with late modeling leads to diverse data sources, mixed structured and unstructured data, and many users. Data warehouses must enable effective data governance to avoid data swamps and data clods, and have robust security mechanisms for data integrity and security.
Challenge 5: The demand for real-time analytical results is increasing, requiring data warehouses to support real-time and streaming data ingestion.
Challenge 6: Analytical systems must run enterprise analytical workloads with guaranteed stability, business continuity, and data integrity. Backup and active-active solutions are needed to ensure data integrity and business continuity.
From the late 20th century to today, distributed analytical databases have evolved over nearly 30 years and are now progressing toward cloud-native data warehouses. They can effectively address the above challenges and meet the future trend of evolving toward data lakehouse architectures.
Building a data warehouse with a distributed analytical database offers advantages: standard, user-friendly SQL, data models suited to common analytical needs across industries, diverse data mining capabilities, easy integration of required data, and convenient access to data and analytical results.
GBase 8a MPP Cluster, GBASE's large-scale distributed parallel database cluster system, has been deployed at scale for over a decade, closely aligned with market needs. It provides basic features valued by the financial industry—virtual clustering, multi-tenancy, rapid online scaling, online node replacement, high availability, and high security—as well as advanced capabilities like active-active clusters, streaming data integration, machine learning algorithm integration, and unified scheduling of heterogeneous computing engines for unified data management.
2. Application Scenarios of Distributed Analytical Databases in the Financial Industry
The financial industry has complex and diverse businesses. Big data analytics scenarios include batch processing, interactive querying, data mining, data lakehouse, and real-time data warehousing. GBase 8a MPP Cluster suitability for each scenario is as follows.
GBase 8a MPP Cluster Application Scenario Proportions
Batch Processing Scenario:
Handles large data volumes, complex business logic, and high resource consumption. Data processing and analysis timeliness is typically T+1, with concurrency of tasks below 100. GBase 8a MPP Cluster's columnar storage engine, optimizer, rich OLAP functions, complex SQL execution optimization, and high-performance data loading make it highly suitable for this scenario.
Interactive Query Scenario:
Requires high query response times for human-machine interaction. GBase 8a MPP Cluster's intelligent indexing, compression, and high-concurrency query support deliver high-performance ad-hoc querying. Its data security mechanisms—encryption, data masking, and access control—safeguard interactive query scenarios.
Data Mining Scenario:
Leverages cross-disciplinary techniques from statistics, machine learning, and database systems to uncover hidden patterns in large datasets. GBase 8a MPP Cluster provides machine learning algorithms for in-depth analysis and mining of user data, turning data into user value.
Data Lakehouse Scenario:
The data lakehouse is a modern, open architecture that bridges data warehouses and data lakes, combining the high performance and management capabilities of a warehouse with the flexibility of a lake. It supports coexistence of multiple data types with data sharing. Through unified encapsulated interfaces, it enables concurrent real-time querying and analytics, bringing greater convenience to enterprise data governance. GBase 8a MPP Cluster offers multiple integration options with the Hadoop platform: through heterogeneous engine integration, it provides a unified access interface, unified data view, and unified computing scheduling; through bidirectional HDFS data export and loading, it enables high-speed data flow between the data lake and data warehouse.
Real-Time Data Warehousing Scenario:
Real-time data warehousing requires real-time data ingestion. Unlike offline warehouses with T+1 response, real-time warehouses demand high real-time responsiveness (minutes down to seconds). Typical applications include real-time OLAP analysis, live dashboards, real-time business monitoring, and real-time data API services. GBase 8a MPP Cluster supports micro-batch data loading and real-time ingestion via Kafka message streams, meeting data warehousing and analysis needs ranging from hourly down to second-level latency.