FaaSConf: QoS-aware Hybrid Resources Configuration for Serverless Workflows
The workflow composition of multiple short-lived functions has emerged as a prominent pattern in Function-as-a-Service (FaaS), exposing a considerable resources configuration challenge compared to individual independent serverless functions. This challenge unfolds in two ways. Firstly, serverless workflows frequently encounter dynamic and concurrent user workloads, increasing the risk of QoS violations. Secondly, the performance of a function can be affected by the resource re-provision of other functions within the workflow. With the popularity of the mode of concurrent processing in one single instance, concurrency limit as a critical configuration parameter imposes restrictions on the capacity of requests per instance. In this study, we present FaaSConf, a QoS-aware hybrid resource configuration approach that uses multi-agent reinforcement learning (MARL) to configure hybrid resources, including hardware resources and concurrency, thereby ensuring end-to-end QoS while minimizing resource costs. To enhance decision-making, we employ an attention technique in MARL to capture the complex performance dependencies between functions. We further propose a safe exploration strategy to mitigate QoS violations, resulting in a safer and efficient configuration exploration. The experimental results demonstrate that FaaSConf outperforms state-of-the-art approaches significantly. On average, it achieves a 26.5% cost reduction while exhibiting robustness to dynamic load changes.
Thu 31 OctDisplayed time zone: Pacific Time (US & Canada) change
10:30 - 12:00 | Cloud and Systems Research Papers / Journal-first Papers / Industry Showcase at Carr Chair(s): Amel Bennaceur The Open University, UK | ||
10:30 15mTalk | FaaSConf: QoS-aware Hybrid Resources Configuration for Serverless Workflows Research Papers Yilun Wang Anhui University, Pengfei Chen Sun Yat-sen University, Hui Dou Anhui University, Yiwen Zhang Anhui University, Guangba Yu Sun Yat-sen University, Zilong He Sun Yat-sen University, Haiyu Huang Sun Yat-sen University Pre-print | ||
10:45 15mTalk | Challenges & Opportunities in Automating DBMS: A Qualitative Study Industry Showcase Yifan WANG Orange/ INRIA, Pierre Bourhis University of Lille, Inria, CRIStAL UMR CNRS 9189, Romain Rouvoy University Lille 1 and INRIA, Patrick Royer Orange | ||
11:00 15mTalk | Test-suite-guided discovery of least privilege for cloud infrastructure as code Journal-first Papers DOI | ||
11:15 15mTalk | Microservice Decomposition Techniques: An Independent Tool Comparison Research Papers Yingying Wang University of British Columbia, Sarah Bornais The University of British Columbia, Julia Rubin The University of British Columbia Pre-print | ||
11:30 10mTalk | Towards Long-Term Scientific Model Sustainment at Sandia National Laboratories Industry Showcase Christian Gilbertson Sandia National Labs, Reed Milewicz Sandia National Laboratories, Eric Berquist Sandia National Labs, Aaron Brundage Sandia National Labs, John Engelmann Sandia National Labs, Brian Evans Sandia National Labs, Nicholas Francis Sandia National Labs, Ernest Friedman-Hill Sandia National Labs, Samuel Grayson Sandia National Labs, Evan Harvey Sandia National Labs, Eric Ho Sandia National Labs, Edward Hoffman Sandia National Labs, Kevin Irick Sandia National Labs, Anagha Krishna Sandia National Labs, Aaron Moreno Sandia National Labs, Joshua Teves Sandia National Labs | ||
11:40 10mTalk | Cloud Resource Protection via Automated Security Property Reasoning Industry Showcase Zhixing Xu Amazon Web Services, Shengjian Guo Amazon Web Services, Oksana Tkachuk Amazon Web Services, Saeed Nejati Amazon Web Services, Niloofar Razavi Amazon Web Services, George Argyros Amazon Web Services |