2021-11-12|

Luo Shutian, a PhD student from the joint laboratory of Human-Machine Intelligent Collaboration System in the Guangdong-Hong Kong-Macao Greater Bay Area and the student of the joint training programme between the University of Macau and the Shenzhen Institute of Advanced Technology of the Chinese Academy of Sciences, Prof Xu Cheng-Zhong, Dean of the Faculty of Science and Technology and Xu Huanle, an Assistant Professor in the Centre for Artificial Intelligence and Robotics, won the Best Paper Award in the 12th ACM Symposium on Cloud Computing (SoCC, 2021). They are the only scholar who has won the award in China (including Hong Kong and Macau) so far. Other accepted papers are from world-renowned universities and research institutions, such as Stanford University, University of California, Berkeley, Princeton University, Microsoft Research, etc.

Titled “Characterizing Microservice Dependency and Performance: Alibaba Trace Analysis“,  the paper is the first to conduct a comprehensive and in-depth analysis of microservice performance and call dependencies in cloud native systems, which has drawn a lot of attention from both academia and industry companies. This paper characterized more than 20,000 microservices and 10 billion microservice call graphs on their dependencies and dynamics. This characterization can be leveraged to optimize microservice system design and improve the quality of online services. In addition, this paper quantified the most important factors that have impact on microservice performance, which could be used to redesign resource management system for improving resource utilization of production clusters. Moreover, this paper built a probabilistic model generator to synthesis  microservice traces in a large scale with accuracy as high as 90%. These studies shall promote further innovations of cloud native infrastructures. In the meanwhile, this paper has open sourced microservice traces from Alibaba production clusters, which can help the research community to better understand real challenges faced by large-scale cloud-native systems.

The ACM Symposium on Cloud Computing (SoCC) has been held since 2010 and sponsored by many world-renowned technology companies such as Google, Microsoft and Amazon. The acceptance rate is approximately 21%. The conference this year attracted more than 160 experts to participate. As the demand for cloud computing continues to expand, the conference has now developed into the top conference in the field of cloud computing. It successfully attracts the participation of many elite scholars in global databases and computer systems, and jointly promoting the research and development of cloud computing technology.

 

粤港澳人機智能協同系统聯合實驗室及澳門大學與中國科學院深圳先進技術研究院聯合培養的博士生羅樹添、科技學院院長須成忠及人工智能與機器人研究中心助理教授徐歡樂在日前舉辦的第十二屆國際計算機協會雲計算研討會 (ACM Symposium on Cloud Computing, SoCC, 2021) 中)唯一奪最佳論文獎,是迄今為止中國(包括港澳地區)唯一獲獎的學者。獲取錄的論文均來自全球知名大學和研究機構,如斯坦福大學、加州大學伯克利分校、普林斯頓大學、微軟研究院等。

羅樹添、須成忠教授及徐歡樂教授的獲獎論文《描述微服務依賴和性能: 阿里巴巴溯源分析》(Characterizing Microservice Dependency and Performance: Alibaba Trace Analysis) ,對大型雲原生系統微服務的性能和調用依賴關係進行全面特徵化分析。是首次有學者對此展開研究,成功引起了全球學術界和工業界的廣泛關注。 論文以阿里巴巴為對象,共分析了2萬多個微服務及100億微服務調用拓撲圖的複雜性和動態性以及對微服務之間的調用依賴進行深入分析,例如用於優化微服務的性能、提高微服務的服務品質等。此外,論文分析了微服務的性能影響因素,以引導未來微服務的資源管理和提升生產集群的資源利用率。文章還構建了微服務調用拓撲圖的概率模型生成器,用於自動化生成大規模的微服務調用拓撲軌跡,其準確率高達90%。 研究結果對雲計算技術領域整體演進產生了深遠價值和影響。論文更開展了阿裡巴巴生產集群中微服務相關數據,方便日後更多研究人員深入探討大規模雲原生系統所面臨的難題和挑戰。

國際計算機協會雲計算研討會(ACM Symposium on Cloud Computing, SoCC) 自2010年開辦,由谷歌、微軟和亞馬遜等眾多全球知名科技公司贊助。錄取率約為 21%。是次會議共有吸引160多名專家參與。隨著社會對雲計算的需要不斷擴大,研討會現已發展成為雲計算領域的頂尖會議,成功吸引全球數據庫和計算機系統眾多精英學者參與,一同推進雲計算技術的研究與發展。

 

澳大科技學院學者奪唯一最佳論文獎

UM FST Scholars achieved the best paper award