Reducing Computational Waste in Configuration Management Through a Lazy Loading and Delta Driven Approach

• Caio R. Oliveira UnB https://orcid.org/0009-0006-1078-7734
• Arthur A. Melo UnB https://orcid.org/0009-0006-1440-1405
• Carla Rocha UnB https://orcid.org/0000-0003-3102-5166
• Isaque Alves USP https://orcid.org/0000-0002-7254-0378

DOI: https://doi.org/10.5753/semish.2026.23797

Resumo

Cloud-scale infrastructure has made Green Computing a major concern in software engineering. Continuous provisioning and management of parallel systems impose a direct, quantifiable environmental footprint. Contemporary agentless Configuration Management (CM) frameworks are designed around idempotency and bulk-fetching optimizations. However, they exhibit latent hardware-level inefficiencies due to the broad use of eager-loading initialization strategies. This approach requires loading of modules and dependencies at initialization, regardless of execution context. As a result, there are redundant I/O operations, unnecessary system calls, and increased CPU idle-wait states. Altogether, these factors contribute to unsustainable energy consumption in continuous delivery pipelines. To overcome this limitation, this study uses the Design Science Research (DSR) methodology to design and empirically evaluate Ch-aOS. Ch-aOS is a novel CM framework that implements a lazy-evaluation architecture via dynamic import dispatch and a delta-driven inmemory approach to idempotency. Modules are loaded strictly on demand at runtime, eliminating superfluous initialization overhead. This limits resource consumption to each execution context’s operational requirements while maintaining batch collection mechanisms needed for idempotency. Experimental findings show that micro-level optimizations, such as reduced system calls and idle-wait CPU time, produce measurable macro-environmental benefits across large-scale deployments. This reduces hardware-level inefficiencies, specifically CPU idle-wait and unnecessary system calls, which are fundamental drivers of energy consumption at the node level.

Referências

Ailane, M. T., Rubner, C., and Rausch, A. (2025). Green devops: A strategic framework for sustainable software development. In Computer Sciences & Mathematics Forum, volume 10, page 5. MDPI.

Angelaki, E., Garefalakis, A., Kourgiantakis, M., Sitzimis, I., and Passas, I. (2025). Esg integration and green computing: A 20-year bibliometric analysis. Sustainability, 17(7):3266.

Bâra, R.-M., Boiangiu, C. A., and Tudose, C. (2024). Analysing the performance impacts of lazy loading in web applications. Journal of Information Systems & Operations Management, 18(1):1–15.

Benson, J. O., Prevost, J. J., and Rad, P. (2016). Survey of automated software deployment for computational and engineering research. In 2016 IEEE SysCon, pages 1–6.

Bucher, B. (2025). PEP 810 – Explicit lazy imports. Guimarães, B. C. F. and Pereira, F. M. Q. (2023). Lazy evaluation for the lazy: Automatically transforming call-by-value into call-by-need. Proc. 32nd ACM SIGPLAN CC.

Gurijala, S. (2025). Sustainable infrastructure: How declarative and immutable systems reduce waste in cloud operations. Intl. Journal of Computational and Experimental Science and Engineering, 11(3):6332–6338.

Hevner, A. R., March, S. T., Park, J., and Ram, S. (2004). Design science in information systems research. MIS quarterly, pages 75–105.

Jin, J. N., Ji, E. K., and Zhang, Q. (2025). Green software engineering: A study on energy-efficient design and deployment in cloud infrastructure. Journal of Data Analysis and Information Processing, 13:241–254.

Kosbar, S. and Hamdaqa, M. (2025). Smells-sus: Sustainability smells in iac. In 2025 IEEE/ACM 22nd MSR, pages 801–812.

Onoja, M. O., Onyenze, C. C., and Akintoye, A. A. (2024). Devops and sustainable software engineering: Bridging speed, reliability, and environmental responsibility. Intl. Journal of Technology, Management and Humanities, 10(04):60–83.

Syed, A. A. M. and Anazagasty, E. (2023). Ansible vs. terraform: A comparative study on iac efficiency in enterprise it. Intl. Journal of Emerging Trends in CS and IT, 4(2):37–48.

Truong, N. K., Pape, C., Rieger, S., and Reißmann, S. (2017). Energy-efficient live migration of i/o-intensive virtual network services across distributed cloud infrastructures. ICSNC 2017, page 13.