{"id":29459,"date":"2025-09-09T00:30:58","date_gmt":"2025-09-09T07:30:58","guid":{"rendered":"https:\/\/www.pingcap.com\/?post_type=case-study&p=29459"},"modified":"2025-09-19T06:04:07","modified_gmt":"2025-09-19T13:04:07","slug":"webank-cuts-costs-scaling-tidb-petabytes","status":"publish","type":"case-study","link":"https:\/\/www.pingcap.com\/ko\/case-study\/webank-cuts-costs-scaling-tidb-petabytes\/","title":{"rendered":"WeBank Cuts Costs by 30% While Scaling TiDB to Petabyte-Level Operations"},"content":{"rendered":"

WeBank, China\u2019s first digital-only bank, delivers inclusive financial services at scale, supporting millions of individual and SME customers across lending, deposits, risk control, and wealth management. Backed by Tencent and founded in 2014, the Shenzhen-based bank serves over 300 million users. With a fully cloud-native architecture and strong commitment to open-source innovation, WeBank has become one of the country\u2019s most technologically advanced financial institutions.<\/p>\n\n\n\n

To meet its demanding performance and reliability needs, WeBank adopted TiDB in 2019 as a foundational layer for mission-critical services. Today, TiDB powers over 80 clusters across multiple business lines, supporting petabytes of data and high-concurrency workloads. Its largest cluster handles over 200TB and 237K QPS, enabling real-time, data-driven operations across the bank.<\/p>\n\n\n\n

This is an excerpt from a live sharing session by Mr. Huang Wei from WeBank\u2019s database team during the TiDB Community event in Shenzhen.<\/p>\n\n\n\n

<\/span>The Challenge: Scaling Beyond the Limits of Legacy Architecture<\/span><\/h2>\n\n\n\n

As WeBank\u2019s customer base and product offerings expanded, the limitations of its traditional database systems became increasingly apparent. The bank\u2019s existing TDSQL architecture<\/a> was built on single-instance, vertically scaled databases. It was well-suited for small-scale transactional workloads but struggled with capacity expansion, concurrency, and cost efficiency at scale. Scaling up often required disruptive, manual migrations and significant re-engineering efforts. Operations teams had to rely on disparate tools to troubleshoot issues, which made root cause analysis<\/a> slow and reactive.\u00a0<\/p>\n\n\n\n

At the same time, the bank set ambitious availability goals. These included 99.999% uptime, zero data loss (RPO=0), and recovery times<\/a> measured in seconds, which their current stack could not guarantee. WeBank needed a future-proof solution that could deliver performance, reliability, and manageability across distributed systems. Especially for mission-critical use cases like core accounting, batch archiving, and compliance reporting.<\/p>\n\n\n\n

Original Data Architecture<\/strong><\/h3>\n\n\n\n

Prior to its TiDB deployment, WeBank used TDSQL for most business workloads, particularly those involving under 3TB of data. These were handled through a single-instance primary-secondary architecture. While stable and simple, this architecture lacked the flexibility to support fast-growing and data-intensive services. TiDB had already been deployed in select batch processing<\/a> and data archiving scenarios but lacked integrated monitoring, diagnostic tools, and centralized control. There was no unified platform to manage large-scale distributed databases. Making it challenging to scale without increasing operational risk or overhead.<\/p>\n\n\n\n

<\/span>The Solution: A Distributed, Intelligent Data Platform for Financial-Grade Operations<\/strong><\/span><\/h2>\n\n\n\n

To address these challenges, WeBank adopted a dual-engine database architecture, combining TDSQL for smaller, simpler workloads and TiDB for high-volume, high-concurrency systems. TiDB\u2019s \uc218\ud3c9\uc801 \ud655\uc7a5\uc131<\/a> made it ideal for supporting data-intensive services like financial accounting, operational analytics, and risk management. Over a six-year period, WeBank expanded its TiDB deployment from 20 to over 80 clusters, managing more than 1.3 petabytes of data across nearly 1,000 servers. To ensure control and visibility at this scale, the database team built an in-house operations platform that integrates cluster management, slow query tracking<\/a>, real-time diagnostics, and capacity forecasting.<\/p>\n\n\n\n

\"\"<\/figure>\n\n\n\n

Fig. 1: Dual engine database architecture using TDSQL and TiDB<\/em><\/p>\n\n\n\n

The team introduced a capacity center that leverages predictive algorithms to estimate future storage needs and growth trends. This allowed them to plan expansions proactively while optimizing resource usage. They also built an intelligent diagnostics engine that automatically gathers logs, metrics, and performance snapshots when alerts are triggered. It delivers root cause analysis directly to the operations team without manual intervention.\u00a0<\/p>\n\n\n\n

Slow query logs were collected and aggregated in real time using an ELK-based stack, enabling immediate insights and anomaly detection<\/a>. More than 50 automated inspection tasks were deployed to monitor everything from replication and backup health to security configurations and capacity thresholds. Together, these tools transformed the way WeBank manages distributed systems. It reduced incident response times, minimized false alarms, and enabled faster, more informed decision-making.<\/p>\n\n\n\n

Solution Data Architecture<\/strong><\/h3>\n\n\n\n

The new solution architecture centered around TiDB as the core platform for hybrid transactional and analytical workloads. TDSQL remained in use for lightweight systems, while TiDB handled large-scale services with varying degrees of complexity and concurrency. Prometheus and ELK were tightly integrated to provide end-to-end observability. TiDB\u2019s compatibility with the MySQL protocol ensured a smooth migration path, and DM (Data Migration) was used to synchronize critical data across systems with strong consistency guarantees. This new architecture allowed WeBank to meet its operational and compliance requirements while delivering the performance and flexibility needed for continued innovation.<\/p>\n\n\n\n

<\/span>Results: Resilience, Speed, and Simplicity at Scale<\/strong><\/span><\/h2>\n\n\n\n

The results of WeBank\u2019s TiDB implementation speak not only to technical performance, but to business transformation. With TiDB at the core of its distributed operations platform, WeBank has achieved the scale, reliability, and agility necessary to meet the demands of a digital-first, regulation-heavy banking environment.<\/p>\n\n\n\n