Landlord: Coordinating Dynamic Software Environments to Reduce Container Sprawl

作     者：Shaffer, Tim Phung, Thanh Son Chard, Kyle Thain, Douglas 

作者机构：Univ Notre Dame Notre Dame IN 46556 USA Univ Chicago Chicago IL 60637 USA 

出 版 物：《IEEE TRANSACTIONS ON PARALLEL AND DISTRIBUTED SYSTEMS》 (IEEE Trans Parallel Distrib Syst)

年 卷 期：2023年第34卷第5期

页      面：1376-1389页

核心收录：

学科分类：0808[工学-电气工程] 08[工学] 0812[工学-计算机科学与技术（可授工学、理学学位）] 

基　　金：NSF [OAC-1931348] DOE Graduate Computer Science Fellowship 

主　　题：Containers Task analysis Software Python Software algorithms Sensitivity High performance computing Cluster computing containers file systems 

摘      要：Containers provide customizable software environments that are independent from the system on which they are deployed. Online services for task execution must often generate containers on the fly to meet user-generated requests. However, as the number of users grows and container environments are changed and updated over time, there is an explosion in the number of containers that must be managed, despite the fact that there is significant overlap among many of the containers in use. We analyze a trace of container launches on the public Binder service and demonstrate the performance and resource usage issues associated with container sprawl. We present Landlord, an algorithm that coalesces related container environments, and show that it can improve container reuse and reduce the number of container builds required in the Binder trace by 40%. We perform a sensitivity analysis of Landlord using randomized synthetic workloads on a high-energy physics (HEP) software repository and demonstrate that Landlord shows benefits for container management across a wide range of usage patterns. Finally, we compare Landlord to offline clustering, and observe that the continuous churn in software necessitates an online approach.