Benchmarking Software Aging Effects in Container Platforms

As containerization becomes increasingly prevalent in modern software development, understanding the long-term effects of software aging is crucial for ensuring the performance and dependability of containerized services. This article introduces a benchmarking framework designed to systematically evaluate software aging in container platforms. Our approach employs well-defined metrics to characterize aging behaviors under representative workloads and stress conditions. Using this framework, we conduct a comparative study of two widely used container platforms, Docker and Podman, across multiple deployment environments (Debian 12, Ubuntu 22, and Ubuntu 24). The results provide insights into their effectiveness in mitigating the effects of software aging. Our findings reveal distinct aging patterns, with Podman showing greater susceptibility to memory leaks while older Docker versions suffer from more significant CPU degradation.