数据与计算发展前沿 ›› 2022, Vol. 4 ›› Issue (5): 50-59.

CSTR: 32002.14.jfdc.CN10-1649/TP.2022.05.006

doi: 10.11871/jfdc.issn.2096-742X.2022.05.006

• 专刊:东数西算:开启算力经济时代的世纪工程(上) • 上一篇    下一篇

“东数西算”趋势下的气象算力网络和算力服务体系架构

杨昕*(),沈文海   

  1. 国家气象信息中心,北京 100081
  • 收稿日期:2022-07-30 出版日期:2022-10-20 发布日期:2022-10-27
  • 通讯作者: 杨昕
  • 作者简介:杨昕,国家气象信息中心,高级工程师,长期从事高性能计算机系统的建设,大规模气象数据处理和服务系统的分析、设计、研发、维护,主要研究方向为高性能计算、信息系统体系结构、分布式计算等。
    负责论文初稿撰写。
    YANG Xin is a senior engineer at the National Meteorological Information Centre. He has long been engaged in the building of High-Performance Computing Systems for Numerical Weather Prediction applications, and the design and deve-lopment of the large-scale meteorological data processing system. His research interests include High-Performance Computing, information system architecture, and distributed systems.
    In this paper, he is responsible for the paper drafting.
    E-mail: yangxin@cma.gov.cn

Meteorological Computing Power and Service Architecture under “East-West Computing Requirement Transfer”

YANG Xin*(),SHEN Wenhai   

  1. National Meteorological Information Centre, Beijing 100081, China
  • Received:2022-07-30 Online:2022-10-20 Published:2022-10-27
  • Contact: YANG Xin

摘要:

【目的】作为具有高度信息化特性的气象业务,对信息系统有强烈的需求和依赖。先进的有持续发展能力的系统体系结构对系统的建设和业务功能效率的支持起着决定性作用。【方法】“全国一体化大数据中心体系”和“东数西算”战略为信息系统架构的建立展现了一个新的视野,并为各行业数字化转型提供了新的方向和思路。本文首先概述了现代“算力体系结构”及其应具备的核心内容及关键特性,然后进一步分析气象行业个性化的特殊性需求。【结果】通过分析并结合“Cloud Native”等现代云计算技术,构想一个“气象算力体系结构”,描述了气象业务框架的核心功能领域以及“气象算力架构”的可持续发展能力。

关键词: 东数西算, 气象算力体系结构, 算力架构, 云计算, 数据中心

Abstract:

[Objective] Meteorological businesses, which are highly information-dependent, always have very strong and unique requirements for information technology. The modern and sustainable system architecture has a decisive impact on the business efficiency and the building of the system. [Methods] The “Country-wide-integrated-grand-data-center” together with the “East-West Computing Requirement Transfer” strategy provide not only a new vision for building the information system architecture, but also a new way of thinking about digital transformation in various industries. Running high-performance computing (HPC) applications such as numerical weather prediction (NWP) usually requires a huge quantity of computing power, while at the same time consuming a large amount of energy. Establishing data centers in the Western part of China can effectively reduce the costs of energy consumption, and consequently results in the reduction of carbon emission. Transmitting data from data centers in the east to data centers in the west and then performing the computing and processing in the west is an ideal way for running data-intensive as well as HPC applications in terms of energy and economy. In this article, we briefly describe modern computing power architecture, its primary contents, and essential core characteristics. We further analyze the customized requirements specific to the meteorology industry. [Results] Then we propose a meteorological computing power architecture based on the above analysis and the state-of-the-art cloud computing technologies, particularly cloud native technologies. This architecture consists of fundamental business frameworks and functionalities in the meteorology industry. The sustainability of this architecture is emphasized.

Key words: East-West Computing Requirement Transfer, meteorological computing power architecture, computing power architecture, cloud computing, data center