Background Introduction
In 1987, China Merchants Bank (CMB) was founded in Shekou Industrial Zone of Shenzhen. It was a pilot bank, the first ever promoted as part of China's reform of the banking industry, and the first non-government bank. CMB is listed on the Shenzhen and Hong Kong Stock Exchanges. Its business covers commercial banking, financial leasing, investment management, life insurance, overseas investment banking, and other financial licenses of the banking group. In recent years, CMB has been focused on the "light-weight bank" strategy, balancing the development of quality, efficiency, and scale. In 2020, CMB started the construction of a comprehensive data middle platform and proposed the idea "everyone is a data analyst".
CMB used Huawei Cloud GaussDB(DWS) to build a next-generation cloud data warehouse with more compute and more storage, improving retail experience with real-time monitoring and data visualization.
In the Cloud Data Warehouse Construction and Joint Innovation project, CMB worked with Huawei to develop a next-generation distributed cloud data warehouse. A 240-node ultra-large cluster was built to process all the retail data applications of the bank and can store up to 10 PB of data, more than four times the storage before. The time required for batch processing tasks has been significantly reduced. The cluster can be easily and inexpensively scaled out. The joint innovation lab established by CMB and Huawei made breakthroughs in backup and recovery, fine-grained disaster recovery, high-speed connections between clusters, and fast warm backup in financial scenarios. The backup speed can reach 150 TB per hour, and the unified logical access of multiple clusters has greatly improved production efficiency.
We needed a large, fast, stable data warehouse, where performance could be massively upgraded by scaling up. The data warehouse cluster should be able to manage thousands of nodes and 100 PB of storage. Regarding its speed, we wanted it to run fully parallel tasks across servers. Stability is also important. We needed multi-active, backup, and multi-level protection to achieve high availability and build a disaster recovery system that could cope with a range of faults. Huawei GaussDB(DWS) could meet all these requirements and was chosen as our next-generation platform.
We needed a large, fast, stable data warehouse, where performance could be massively upgraded by scaling up. The data warehouse cluster should be able to manage thousands of nodes and 100 PB of storage. Regarding its speed, we wanted it to run fully parallel tasks across servers. Stability is also important. We needed multi-active, backup, and multi-level protection to achieve high availability and build a disaster recovery system that could cope with a range of faults. Huawei GaussDB(DWS) could meet all these requirements and was chosen as our next-generation platform.