Information
Center
资讯中心
Chinese Company
中国企业
Ant Group: High-quality and Low-carbon Development Practices Based on Green Computing Technology | Case Presentation of "2024 'Beautiful China, I Am an Actor' Enterprise Climate Action Case Collection" (Case No. 9)
蚂蚁集团:基于绿色计算技术的高质量、低碳发展实践 | 《2024“美丽中国,我是行动者”企业气候行动案例集》案例展示(九)
2025-03-12
来源:微言环保
作者:微言环保
Editor's Note
During the COP29 conference, the Publicity and Education Center of the Ministry of Ecology and Environment, the Da Ya Center for Climate Change Mitigation, and the Vanke Charity Foundation jointly released the "2024 'Beautiful China, I Am an Actor' Enterprise Climate Action Case Collection" with 15 selected cases showcasing the active role and innovative practices of Chinese enterprises in ecological civilization construction.

Over the past six years, a total of 95 enterprises' low-carbon transformation cases have been included in the annual "Enterprise Climate Action Case Collection" and were released at the United Nations Climate Conference, demonstrating the pioneering actions and exemplary power of Chinese enterprises in responding to climate change.
We present here the selected cases of the "2024 'Beautiful China, I Am an Actor' Enterprise Climate Action Case Collection", expecting these cases to inspire and drive more enterprises to join the innovative emission reduction actions, and jointly promote the realization of the "dual carbon" goals.

Ant Technology Group Co., Ltd.
High-quality and low-carbon development practices based on green computing technology

Expert Comment:
Ant Group has released the carbon neutrality target roadmap for 2030 net zero emissions, demonstrating its firm commitment as an industry leader in green and low-carbon development. The innovation of Ant Group in green computing technology not only effectively solved industry problems such as large-scale cluster resource allocation and intelligent traffic prediction, but also significantly improved resource utilization, setting an industry benchmark for data center energy conservation and emission reduction. Especially in the large-scale application in high-traffic scenarios like "Double Eleven", it proved the practical feasibility and efficiency of green computing technology. Through open-source green computing technology, Ant Group has built an open and win-win technology cooperation ecosystem, lowering the threshold for industry application of green computing and making significant contributions to promoting the sustainable development of the entire society.



01
Case Overview

Based on the concepts and principles of sustainable development and in accordance with the ESG framework, Ant Group actively responded to the national carbon neutrality goals in 2021 and released a carbon neutrality target roadmap, committing to achieving net zero emissions (Scope 1, 2, and 3) by 2030.

During the process of practicing green and low-carbon development, Ant Group has consistently played three roles: 1) Practitioner of green development for enterprises: Committing to and fulfilling the carbon neutrality goals; 2) Provider of green and low-carbon products and services: Guiding the public to participate in green living and ecological protection and restoration to form a cycle; 3) Explorer of green technologies: Promoting the green transformation of industries.

Ant Group conducts green computing technology innovation and practices the exploration of green technologies, focusing on core technologies such as low-carbon computing power selection capabilities, global resource scheduling capabilities, workload optimization capabilities, and green monitoring and evaluation capabilities. These technologies are applied to Ant Group's server clusters.

In 2021, green computing technology was scaled up and applied to the "Double Eleven" business scenarios, and continuous iterations were carried out from 2021 to 2024. Currently, it has been applied on a large scale in Ant Group's daily operations and during the "Double Eleven" peak period. It effectively solved industry problems such as rational allocation of large-scale cluster resources, minute-level effective scheduling, and intelligent traffic prediction, significantly improving resource utilization and demonstrating a leading effect in energy conservation and emission reduction for data centers in various industries.

In 2023, Ant Group reduced carbon emissions from upstream data centers in its supply chain by 72,026.70 tons of carbon dioxide equivalent through green computing technology. Compared to the baseline usage scenario with the same server scale, using green computing technology can reduce carbon emissions by up to 26%.



02
Company Profile

Ant Group originated from Alipay, which was established in 2004. It was born out of the original intention of solving trust issues for society. After nineteen years of development, it has become the world's leading Internet open platform.
We leverage technological innovation to assist our partners, providing inclusive and convenient digital life and digital financial services for consumers and small and micro enterprises; we continuously open up products and technologies to help enterprises achieve digital transformation and collaboration;
We collaborate extensively globally, serving local merchants and consumers to facilitate "global payment", "global settlement", and "global remittance".


03
Project Outcomes

1. Industrial Benefits:
(1) Current "green computing power" in the industry usually focuses on hardware and is centered around power generation and computing supply. The implementation of green computing in Ant Group as a typical practice of "green computing in the computing application环节" will promote the industry's understanding and development of end-to-end green computing.

2. Economic Benefits
(1) Internal Application: Direct Economic Benefits
Take Alipay as an example. The upgrade capability of the entire station's infrastructure middleware has been shortened from an annual cycle to a bi-weekly cycle. In terms of operation efficiency, it is possible to achieve unmanned 15-day upgrades for tens of thousands of business services across the entire station, significantly improving R&D and operation efficiency and reducing the cost of thousands of people, saving the company tens of billions of yuan annually.
(2) Potential Promotion: Indirect Economic Benefits
Looking forward, if green computing is promoted in the financial industry, annual electricity savings of 4.5 billion yuan can be achieved by improving the utilization rate of computing resources, reducing carbon emissions by 160,000 tons, and potentially saving the purchase of 100,000 servers, generating 7 billion economic benefits. There are also potential promotion opportunities in the Internet and other industries.
3. Social Benefits
(1) Supporting ESG Goals of Enterprises and National Carbon Neutrality Strategy
In 2023, Ant Group reduced carbon emissions from upstream data centers in the supply chain by 72,026.70 tons of carbon dioxide equivalent through green computing technology. This not only helped enterprises achieve their ESG goals but also contributed to the national carbon neutrality strategy.
(2) Building an Open and Win-Win Technological Cooperation Ecosystem through Open Source
Ant Group has currently committed to opening up 7 low-carbon patents externally and will gradually open the core software involved in the green computing technology system to the industry, reducing the practical threshold for green computing applications in the industry. 04
Project Highlights

- Evaluated by the National Energy Conservation Center, Ant Green Computing, as a green and low-carbon technology, has reached the international leading level.
- One of the 2022 Annual Dualization and Synergy Typical Cases of Digital Technology Enterprises organized by the Cyberspace Administration of China.
- Selected for the 2023 List of Typical Cases of Digital and Green Transformational Development in the Yangtze River Delta (Shanghai) jointly organized by the Cyberspace Administration Offices of the Three Provinces and One Municipality in the Yangtze River Delta.
- 2023 36 Carbon Zero Carbon Pioneer Case, etc.


05
Project Implementation

Ant Group, as a typical representative of internet enterprises, has servers ranking among the top ten in the world. Since 2018, the number of servers within Ant Group has doubled year by year. At that time, the utilization level of servers in the industry was very low. A large amount of energy was consumed in maintaining servers instead of providing real computing power. In 2017, the overall server utilization rate of Ant was only about 8%, and a large number of servers were in idle state. Therefore, Ant Group launched the "Cloud Connect Future" project. One of the key goals was to improve the overall operational efficiency of Ant's data center through technological architecture upgrades by rationally allocating computing resources, thereby achieving green and stable development.
Ant Group has established a green computing system. On the premise of using green data centers, through the capabilities of its own high-quality software stack, including the upgrade of computing platform system architecture, business software updates, and the collaborative design of business applications and platform systems, it improves the efficiency of computing resource utilization. This "green computing" technology system focuses on resource efficiency improvement and is centered around low-carbon computing power selection capabilities, global resource scheduling capabilities, workload optimization capabilities, and green monitoring and evaluation capabilities. It has now been widely applied in Ant Group's business scenarios.
The Ant Group's green computing technology system includes multiple technical systems such as hardware, operating systems, efficient scheduling systems, and multi-dimensional intelligent data analysis. The key technical path of this system is:
1. Through pooling and other technical means, integrate heterogeneous server computing power to provide unified computing power services in a standardized way, isolate the complexity of heterogeneous computing power, and improve stability;
2. Through elastic computing power unified scheduling, performance optimization, and containerization to improve system performance; reduce equipment dependence;
3. Through practical platform engineering and code-based configuration to efficiently and rapidly perform resource elastic scaling, improve operation efficiency and support capacity for peak business activities;
4. Through Serverlessization, make workloads more easily elastic scalable and schedulable;
5. Workload optimization: In combination with Ant Group's own business needs, focus on optimizing storage workloads, AI intelligent computing workloads, and online workloads;
6. Through the green observation and optimization system, continuously detect workload resource consumption and optimize resources.

This project has achieved large-scale application of green computing technology in Ant Group's daily operations and during the Double Eleven peak period. It has stably supported the core business systems of Netcom Bank and Alipay, ensuring the stable operation of related systems for over four years. It has effectively solved industry problems such as rational allocation of large-scale cluster resources, minute-level effective scheduling, and intelligent traffic prediction, significantly improving resource utilization, and has demonstrated a demonstration effect for energy conservation and emission reduction in data centers of various industries.
In 2023, Ant Group reduced carbon emissions from upstream data centers in its supply chain by 72,026.70 tons of carbon dioxide equivalent through green computing technology. Compared to the benchmark usage scenario with the same server scale, using green computing technology can reduce carbon emissions by up to 26%. 06
Project Impact, Usability and Sustainability

Key Innovations:
1. By leveraging a unified scheduling capability encompassing offline hybrid deployment technology, time-based scheduling technology, AI intelligent capacity technology, and Serverless (serverless) architecture, the utilization rate of computing resources has been significantly enhanced.
2. Through Continuous Profiling (continuous performance analysis), resource bottlenecks and root causes can be quickly identified, and the GreenOps platform has been pioneered in the industry, focusing on the carbon emission range III of data centers, providing users with a one-stop capability for carbon emission observation, analysis, and optimization. It has now been piloted and implemented in multiple internal business areas, significantly improving green observability.
Green computing technology has the potential for scalability, including:
- Industry-wide applicability of business scenarios: Green computing technology can be applied to enterprises within industries such as internet and finance that have refined demands for improving cloud resource utilization and reducing carbon emissions.
- Hardware universality: The hardware in the green computing system is entirely based on general-purpose hardware and domestic hardware, which can be selected as an option or scaled out through domestic server manufacturers.
- Standardization of technology/products: The core software and key technologies in the green computing system are gradually improving their standardization through open source and commercial forms.
Future Plans:
Looking to the future, Ant Group will actively participate in the establishment of industry standards and evaluation systems, collaborate with multiple parties including academia, industry, and research institutions to jointly build a green computing ecosystem. At the same time, it will also actively open up green computing practices and export them externally through open source and commercial forms.

编者按

在COP29会议期间,生态环境部宣传教育中心、大道应对气候变化促进中心联合万科公益基金会,共同发布了《2024“美丽中国,我是行动者”企业气候行动案例集》,15家入选案例展示了中国企业在生态文明建设中的积极角色和创新实践。



过去6年,共有95家企业的低碳转型案例被收录于年度《企业气候行动案例集》并在联合国气候大会上发布,展现了中国企业应对气候变化的先锋行动和榜样力量。


我们在此呈现《2024“美丽中国,我是行动者”企业气候行动案例集》的入选案例,期待这些案例可以激励和推动更多企业加入创新减排的行动中,共同推动“双碳”目标的实现。


蚂蚁科技集团股份有限公司

基于绿色计算技术的高质量、低碳发展实践


专家点评:


蚂蚁集团通过发布 2030 年净零排放的碳中和目标路线图,展现了其作为行业领导者在绿色低碳发展方面的坚定承诺。蚂蚁集团在绿色计算技术上的创新,不仅有效解决了大规模集群资源分配、智能流量预测等行业难题,还显著提升了资源利用率,为数据中心节能减排树立了行业标杆。特别是在“双十一”等高流量场景中的规模化应用,证明了绿色计算技术的实际可行性和高效性。蚂蚁集团通过开源绿色计算技术,构建了开放共赢的技术合作生态,降低了行业应用绿色计算的门槛,为推动全社会的可持续发展做出了重要贡献。


01

案例概述

基于可持续发展的理念与原则,结合ESG的框架,蚂蚁集团积极响应国家双碳目标,在2021年发布了碳中和目标路线图,承诺在2030年实现净零排放(范围一、二、三)。


在践行绿色低碳发展的过程中,蚂蚁集团坚持扮演三重角色:1、企业绿色发展的践行者:承诺并践行碳中和目标;2、绿色低碳产品与服务的提供者:引导社会大众参与绿色生活和生态保护与修复构成循环;3、绿色技术的探索者:推动产业绿色转型。


蚂蚁集团进行绿色计算技术创新和实践进行绿色技术的探索,围绕低碳算力选择能力、全局资源调度能力、工作负载优化能力以及绿色监测评估能力等方面进行了一系列核心技术攻关,并将其应用到蚂蚁集团服务器集群。


2021年开始将绿色计算技术规模化应用于“双十一”业务场景,并在2021-2024年持续迭代,目前已实现在蚂蚁集团日常和双十一高峰期间的大规模应用。有效解决了大规模集群资源合理分配、分钟级有效调度、智能流量预测等行业难题,显著提升了资源利用率,为各行业数据中心节能减排起到了示范效应。


2023年,蚂蚁集团通过绿色计算技术共减少供应链上游数据中心碳排放72,026.70吨二氧化碳当量,相较于同等服务器规模的基准使用情景,使用绿色计算技术后,可减少高达26%的碳排放比例。


02

机构简介

蚂蚁集团起步于2004年诞生的支付宝,源于一份为社会解决信任问题的初心,经过十九年的发展,已成为世界领先的互联网开放平台。


我们通过科技创新,助力合作伙伴,为消费者和小微企业提供普惠便捷的数字生活及数字金融服务;持续开放产品与技术,助力企业的数字化升级与协作;


在全球广泛合作,服务当地商家和消费者实现“全球收”、“全球付”和“全球汇”。


03

项目成果

1、产业效益:

(1)行业目前的“绿色算力”通常围绕硬件展开,聚焦在电力生产、算力供给环节,绿色计算在蚂蚁集团的落地作为“算力应用环节的绿色计算”的典型实践,将推动端到端的绿色计算行业认知和发展。


2、经济效益

(1)内部应用:直接经济效益

以支付宝为例,全站基础设施中间件的升级能力从以年为周期缩短到以两周为周期,在运维效率上可以做到15天全站几十万业务服务无人值守升级,显著提升了研发及运维效率,降低数千人耗费,为公司每年节省成本数十亿元。

(2)潜在推广:间接经济效益

展望未来,如在金融业推广绿色计算,每年可通过提升算力资源利用率节省电费45亿元,减碳16万吨,每年或节省10万台服务器采购,产生70亿经济效益。互联网等行业也有潜在的推广机会。


3、社会效益

(1)助力企业ESG目标和国家双碳战略

2023年,蚂蚁集团通过绿色计算技术共减少供应链上游数据中心碳排放72,026.70吨二氧化碳当量。在助力企业ESG目标实现的同时也助力了国家双碳战略。

(2)通过开源构建开放共赢的技术合作生态

蚂蚁集团目前已承诺向外部开放7项低碳专利,并将绿色计算技术体系所涉及到的核心软件逐步开源给行业,降低了行业绿色计算应用的实践门槛。


04

项目亮点

-经国家节能中心评价,蚂蚁绿色计算作为绿色低碳技术,已达到国际领先水平

-网信办组织的2022年度数字科技企业双化协同典型案例

-入选由长三角三省一市党委网信办联合组织的2023年长三角数字化绿色化协同转型发展典型案例名单(上海市)

-2023年36碳零碳先锋案例等


05

项目实施

蚂蚁集团作为互联网企业的典型代表,其服务器规模排名世界前十。从2018年开始,蚂蚁集团内部的服务器数量已经呈现翻番式的增长。当时业内的服务器利用水平很低,大量的能源被消耗在维护服务器上,而没有去提供真正的算力,2017年时,蚂蚁的整个服务器利用率只有8%左右,大量的服务器是处于闲置状态。因此,蚂蚁集团启动了“云通未来”项目,关键目标之一就是通过技术架构的升级,合理分配计算资源,提升蚂蚁数据中心的整体运行效率,从而实现绿色、稳定的发展。


蚂蚁集团构建绿色计算体系,在使用绿色数据中心的前提下,通过蚂蚁本身全套高质量软件栈的能力,包括计算平台系统架构升级、业务软件更新以及业务应用与平台系统协同设计等方法提升算力资源利用效率。这套聚焦于资源提效的“绿色计算”技术体系,围绕低碳算力选择能力、全局资源调度能力、工作负载优化能力和绿色监测评估能力展开,目前已经广泛应用于蚂蚁集团的业务场景当中。


蚂蚁绿色计算技术体系包含硬件、操作系统、高效调度系统、多维度智能数据分析等多个技术体系,该方案的关键技术路径为:

1、通过池化等技术手段把异构服务器算力归一,以标准化的方式提供统一算力服务,隔离异构算力的复杂性,提升稳定性;

2、通过弹性算力统一调度、性能优化等,提升系统性能;通过容器化减少设备依赖;

3、通过实践平台工程和配置代码化高效快速的进行资源弹性伸缩,提高运维效率和业务峰值活动支撑能力;

4、通过Serverless化,让工作负载本身更加容易被弹性伸缩和被调度;

5、工作负载优化:结合蚂蚁自身业务需求,着重对存储工作负载、AI智能计算工作负载以及在线工作负载进行了优化;

6、通过绿色观测和优化体系,持续探测工作负载的资源消耗并进行资源优化。



该项目实现了绿色计算技术在蚂蚁集团日常和双十一高峰期间的大规模应用,已经稳定支撑网商银行、支付宝全部核心业务系统,确保相关系统超过四年的稳定运行。有效解决了大规模集群资源合理分配、分钟级有效调度、智能流量预测等行业难题,显著提升了资源利用率,为各行业数据中心节能减排起到了示范效应。


2023年,蚂蚁集团通过绿色计算技术共减少供应链上游数据中心碳排放72,026.70吨二氧化碳当量,相较于同等服务器规模的基准使用情景,使用绿色计算技术后,可减少高达26%的碳排放比例。


06

项目影响力、可推广性与可持续性

创新点:

1、通过包括在离线混部技术、分时调度技术、AI智能容量技术在内的统一调度能力,以及Serverless无服务器化,大幅提升算力资源利用率;

2、通过Continuous Profiling(持续性能分析)快速定位资源瓶颈及根因分析;并且在业界率先研发探索GreenOps平台,聚焦在碳排范围三的数据中心部分,为用户提供一站式的碳排放观测、分析、优化的能力。目前已在内部多个内部业务上进行试点落地,大幅提升绿色可观测性。


绿色计算技术具备可推广性,包括:

-业务场景的行业通用性:绿色计算技术可适用于互联网、金融等行业内存在精细化提升云资源利用率与降碳诉求的企业

-硬件通用性:绿色计算体系中的硬件完全基于通用硬件以及国产硬件,可以作为选配或通过国产服务器厂商进行规模化输出

-技术/产品标准化:绿色计算体系中的核心软件和关键技术通过开源、商业化等形式逐步完善其标准化程度


后续计划:

面向未来,蚂蚁一方面积极参与行业标准和评测体系共建,与产学研多方合作,共建绿色计算生态;另一方面,也积极开放绿色计算实践,通过开源、商业化等形式对外输出。