A research team led by Xu Cheng-Zhong, chair professor in the Faculty of Science and Technology (FST) of the University of Macau (UM), and Xu Huanle, assistant professor in FST, have made significant breakthroughs in the field of cloud computing. The team has designed an innovative resource management system that can improve the efficiency of computing resource use and reduce CPU resource use by nearly 1.6 times. The research results have been published in ACM Transactions on Computer Systems (ToCS), a leading computer science journal.
The artificial intelligence sector is undergoing rapid technological change. With a soaring demand for computing resources, optimising the efficiency of computing resources and accommodating larger computational workloads have become challenges for the cloud computing industry. To improve efficiency in the use of computing resources, the team has designed a new resource management system called Erms. The system is capable of dynamically deploying resources and optimising resource scaling based on actual workloads, and, for the first time, achieving optimal resource management in large-scale, complex microservice scenarios. In addition, the team has designed a new set of scheduling strategy to optimise resource allocation for shared microservices, which significantly improves the efficiency of resource utilisation. Compared with existing microservice systems, Erms can reduce the likelihood of SLA (service-level agreement) violations to one-fifth of the original risk, and achieve nearly 1.6 times savings in CPU resources.
Titled ‘Optimizing Resource Management for Shared Microservices: A Scalable System Design’, the paper is the first study to fully address microservice multiplexing scenarios. It has been published in ACM Transactions on Computer Systems (ToCS). The journal enjoys a high reputation in the field of computer system and has published many significant research results on computer operating systems, networks, databases, and distributed systems. Since its inception 40 years ago, the journal has included only a limited number of papers from China, and the abovementioned paper is the first contribution from the Guangdong-Hong Kong-Macao Greater Bay Area.
Prof Xu Cheng-Zhong and Prof Xu Huanle are the corresponding authors of the study. Luo Shutian, a doctoral graduate co-trained by UM and the Chinese Academy of Sciences (now a postdoctoral fellow at Yale University), is the first author. The research project was supported by the Science and Technology Development Fund of the Macao SAR (File no: 0024/2022/A1), the Key Research and Development Program of the Ministry of Science and Technology (File no: 2019YFB2102100), and the Guangdong Province Key Research and Development Program (File no: 2020B010164003). The full version of the research article can be viewed at https://doi.org/10.1145/3631607.
Furthermore, Prof Xu Cheng-Zhong’s team was awarded the sole Best Paper award at the ACM Symposium on Cloud Computing 2021 for their paper titled ‘Characterizing Microservice Dependency and Performance: Alibaba Trace Analysis’. This marks the first instance of a scholar from China (including Hong Kong, Macao, and Taiwan) receiving the award since the conference’s establishment in 2009.
Both of the abovementioned papers are the results of the team’s collaboration with the Shenzhen Institute of Advanced Technology under the Chinese Academy of Sciences and Alibaba, an international and leading cloud computing company. The team has been funded by the Alibaba Innovative Research Program for five consecutive years and received the Alibaba Outstanding Collaborative Project Award in 2022. Mei Hong, chairman of the China Computer Federation and director of the Academic Committee of the State Key Laboratory of Internet of Things for Smart City, said that cloud computing plays a key role in the wave of smart cities and artificial intelligence models, adding that the publication of the two important papers indicates that UM’s research in the field of cloud computing has achieved international standing.
澳門大學科技學院講座教授須成忠及助理教授徐歡樂的團隊在雲計算領域取得突破性研究進展。團隊提出了一個創新的資源管理系統，可提高計算資源利用效率，並節約CPU資源近1.6倍。研究成果已發表於計算機頂級期刊 ACM Transactions on Computer Systems（ToCS）。
人工智慧領域正經歷日新月異的科技變革，面對海量的計算資源需求，最大化計算資源的利用效率，支持更多計算量成為雲計算業界的挑戰。為了提高計算資源利用效率，團隊設計了全新資源管理系統Erms，能夠根據實際的工作量對資源進行動態調配和資源伸縮優化，並第一次在大規模複雜微服務場景下實現最優資源管理。此外，團隊還設計了一套新的調度策略，用以優化共用微服務的資源配置，大大提升了資源使用的效率。與現有微服務系統相比，Erms能將SLA違規的可能性降低到原來的五分之一，並節約CPU資源近1.6倍。
上述研究的論文《優化共享微服務的資源管理：可擴展的系統設計》是第一個完整解決微服務複用場景的研究，對雲原生系統的後續深入研究有重大啟發意義，現已在計算機期刊 ACM Transactions on Computer Systems （ToCS）發表。該期刊在計算機系統領域享有崇高聲譽，眾多計算機操作系統、網絡、數據庫和分佈式系統中的重要成果都是在該期刊發表，創刊40年以來收錄極少中國發表的文章，是次發表更是粵港澳大灣區的首篇文章。
論文的通訊作者為須成忠和徐歡樂，澳門大學與中國科學院聯合培養的博士畢業生、現為耶魯大學博士後羅樹添為第一作者。論文由澳門特別行政區科學技術發展基金（檔案編號：0024/2022/A1）、科技部重點研發計劃（編號：2019YFB2102100）和廣東省重點領域研發計劃（編號：2020B010164003）資助。論文的完整版本可瀏覽：https://doi.org/10.1145/3631607。
此外，須成忠研究團隊於2021年在國際計算機協會雲計算頂級會議（ACM Symposium on Cloud Computing ）發表的《描述微服務依賴和性能：阿里巴巴溯源分析》獲該年唯一的最佳論文獎。該獎是會議自2009年舉辦以來，首次由中國學者（含港澳台）獲得。
兩篇論文皆為團隊與中國科學院深圳先進技術研究院和雲計算國際龍頭企業阿里巴巴公司的合作成果，團隊的研究工作連續五年獲阿里巴巴“創新研究計劃”資助，並於2022年獲阿里巴巴優秀項目合作獎。中國計算機學會理事長、智慧城市物聯網國家重點實驗室學術委員會主任梅宏院士認為“在智慧城市和人工智能大模型浪潮中，雲計算起到核心作用，這兩篇重要論文的發表標誌澳大在雲計算方面的研究進入國際領先水準”。
The UM research team develops a cloud computing system that improves computing resource efficiency by 1.6 times
澳大雲計算研究可節約1.6倍計算資源
(from left) Xu Huanle, Xu Cheng-Zhong and Luo Shutian
（左起）徐歡樂、須成忠、羅樹添