This site uses cookies. By continuing to browse the site you are agreeing to our use of cookies. Read our privacy policy>

Если вам нужна помощь, пожалуйста, нажмите здесь:

Huawei Builds Earthquake Monitoring System

Background

The Institute of Disaster Prevention (IDP) was established by the State Seismological Bureau in 1975. With a focus towards the disaster prevention industry, IDP provides core majors such as geophysics, geology, resources exploration engineering, exploration technology and engineering, and groundwater science and engineering. At present, over 8,000 full-time students are studying at IDP.

Challenges

IDP pays specific attention to scientific research and regards this area as an important means for improving teaching quality. In recent years, IDP not only provided scientific and teaching research of various levels but contributed to urban disaster prevention and reduction planning, processing of earthquake precursor monitoring data, prediction of urban earthquake disasters, publicity for emergency handling skills, contingency planning, and scientific research for key earthquake disasters.

To minimize loss of lives and property, it is necessary to forecast earthquakes as soon as possible. Earthquake forecasting relies on precise observation and calculation of earthquake precursors. Observation of earthquake precursors records the process of how an earthquake is formed and analyzes changing features in the process to forecast earthquakes. Therefore, a complete IT infrastructure is required to provide a basic computing environment. The construction of the earthquake precursor monitoring laboratory in this project is a major part of national disaster prevention.

The laboratory receives and analyzes satellite data in real-time and compares the data with historical data, supporting earthquake prediction in early stages and assessment after disasters. The analysis of geographical data before and after disasters provides theoretical bases and technological guidelines for decision-making. High efficiency is required to reduce the processing time for such massive amounts of earthquake data during the process of receiving, analyzing, and comparing data. The landform pictures received from the satellite generate large quantities of data. Therefore, the system must support receiving data at speeds of dozens of Mbit/s when connecting to the satellite for receiving data which also requires a large storage capacity (hundred-TB level). To process various types of data at a time, the system needs to support different speeds for accessing historical and real-time data, as well as provide a large-capacity hierarchical storage system for earthquake data storage and access. The massive number of earthquake data calculations require a high-performance computing cluster with the computing capability at the TFLOPS level. As data is accumulated and software is optimized, the system must support smooth capacity expansion to maximize Return On Investment (ROI).

Solution

IDP requires a system that provides powerful computing capabilities, advanced processing software, and high-speed network transmission. In phase one of implementation, the computing system uses the Huawei RH2288 rack servers and high-performance GPUs, providing up to 60 compute nodes and a computing capability at peak value reaches 20 TFLOPS. The storage system uses two dual-controller Huawei OceanStor S5600T optical storage devices to provide a storage capacity of up to 500 TB. SSDs of 4.8 TB are configured to support hierarchical storage. Four RH2285 V2 rack servers are used as I/O nodes, and one RH2285 V2 rack server is used as a metadata management node, which provides high read and write efficiency. The network system uses the Mellanox 4036 QDR switch and Huawei OceanStor SNS 2120 optical switches for communication, ensuring a high network bandwidth for receiving and processing data. Two 2-socket Huawei RH2288 rack servers and CHESS cluster management software from ClusterTech provide efficient resource management, job scheduling, and cluster management.

Benefits

The Huawei RH2288 rack servers greatly improve overall system performance. The 2-socket RH2288 rack servers provide excellent computing performance and the corresponding capability for HPC. The system offers an overall computing performance of 20 TFLOPS, which improves system performance by 80 percent. This creates a solid hardware foundation for earthquake data processing, offering timely and precise observation results for earthquake precursors. By improving system computing power, HPC expands the overall scientific research capabilities for IDP.

The hierarchical storage system efficiently increases the data access speed. Two dual-controller Huawei OceanStor S5600T optical storage devices provide a storage capacity of up to 500 TB. SSDs with 4.8 TB are configured to support hierarchical storage, improving data accessing performance by 70 percent. This improves requirements for storage and access of large quantities of earthquake data, enabling previously impossible research computations.

With a modular design and integrated delivery solution, the system offers high scalability which meets expansion requirements for the next ten years. IDP is now planning phase two of its implementation as scientific research tasks are increasing. With phase one complete, the system supports smooth capacity expansion, ensuring service continuity and maximizing ROI.