Co-Locating Online Workload and Offline Workload in the Cloud: An Interference Analysis
Wenyan Chen, Kejiang Ye, Chengzhong Xu · 2019
Workload characterization plays an important role in resource allocation and performance optimization. In order to improve the resource utilization, cloud providers like Google and Alibaba usually co-locate online and offline jobs on the same cluster. However, the characteristics of co-located services running in containers is still not clear. In this paper, we explore various characteristics of micro-architecture and application level of co-located workloads running in containers on the same server via two popular monitor tools: Perf and Prometheus. Our study focuses on quantifying interference of workloads by analyzing the characteristics of workloads running alone and co-located. Based on the interference analysis, we can design more efficient container deployment or scheduling algorithms in the future.