Performance Antipatterns: Angel or Devil for Power Consumption?

2026-02-12

AI Summary

This study investigates whether the impact of performance antipatterns on energy consumption in microservice systems aligns with their adverse effects on response time. By injecting ten canonical performance antipatterns under controlled workloads and simultaneously measuring performance metrics, CPU/memory power consumption, and resource utilization, the work reveals—through statistical significance analysis—a phenomenon termed “energy-performance decoupling”: certain antipatterns degrade latency without significantly increasing instantaneous power draw. The research identifies a subset of true energy-intensive antipatterns (e.g., Unnecessary Processing, The Ramp) that substantially elevate energy consumption, while others exhibit saturated CPU usage leading to stable power levels. These findings provide empirical evidence and actionable insights for designing energy-efficient microservice architectures.

Technology Category

• Data Mining & Knowledge Management: Scalability, Parallel & Distributed Systems
• Machine Learning: Scalability of ML Systems
• Search and Optimization: Metareasoning and Metaheuristics

Application Category

• Systems and Infrastructure for Web, Mobile and WoT: Web performance, measurement, and characterization
• User Modeling, Personalization and Recommendation: Attacks and countermeasures in recommendation systems
• Security and Privacy: Large-scale security measurements

Abstract

Performance antipatterns are known to degrade the responsiveness of microservice-based systems, but their impact on energy consumption remains largely unexplored. This paper empirically investigates whether widely studied performance antipatterns defined by Smith and Williams also negatively influence power usage. We implement ten antipatterns as isolated microservices and evaluate them under controlled load conditions, collecting synchronized measurements of performance, CPU and DRAM power consumption, and resource utilization across 30 repeated runs per antipattern. The results show that while all antipatterns degrade performance as expected, only a subset exhibit a statistically significant relationship between response time and increased power consumption. Specifically, several antipatterns reach CPU saturation, capping power draw regardless of rising response time, whereas others (\eg Unnecessary Processing, The Ramp) demonstrate energy-performance coupling indicative of inefficiency. Our results show that, while all injected performance antipatterns increase response time as expected, only a subset also behaves as clear energy antipatterns, with several cases reaching a nearly constant CPU power level where additional slowdowns mainly translate into longer execution time rather than higher instantaneous power consumption. The study provides a systematic foundation for identifying performance antipatterns that also behave as energy antipatterns and offers actionable insights for designing more energy-efficient microservices architectures.

Problem

• performance antipatterns
• energy consumption
• microservices
• power usage
• energy antipatterns

Innovation

• performance antipatterns
• energy consumption
• microservices
• power efficiency
• empirical study