Carbon-aware Storage Tiering Framework
DOI:
https://doi.org/10.4108/aism.14563Keywords:
Tiered storage, cloud storage, workload prediction, hot data management, energy-efficient storage, carbon- aware computing, data migrationAbstract
The rapid growth of cloud services and serverless applications has significantly increased the demand for efficient storage management in modern data centers. Tiered storage architectures, which combine high- performance solid-state drives (SSDs) with high-capacity hard disk drives (HDDs), are commonly used to balance performance and cost. However, accurately identifying which data objects should reside in fast storage tiers remains a challenging task due to dynamic and skewed access patterns in real-world workloads. In response, this paper introduces a predictive and carbon-aware storage management framework designed to improve data placement decisions in tiered storage systems. Our approach analyzes blob access traces and employs workload prediction to estimate future data hotness, enabling proactive migration of frequently accessed objects to high-performance storage tiers. By integrating prediction with migration cost awareness and time-varying grid carbon-intensity signals, the framework defers eligible migrations during carbon- intensive periods and prioritizes them during lower-carbon periods. Evaluation on an activity-focused subset of the Azure Functions Blob Access Trace, together with larger-scale experiments using up to 1,000 blobs, shows that the carbon-aware policy reduces operational carbon emissions by 27.76% relative to an otherwise equivalent LSTM energy-aware policy, while reducing total energy consumption by 24.70% and the number of migrations by 50%. A constant-carbon-intensity counterfactual further shows that temporal carbon-aware scheduling contributes an additional 3.05 percentage points of emissions reduction beyond the reduction attributable to lower energy consumption alone. The migration-energy-weighted carbon intensity is reduced from 269.66 to 174.65 gCO2e/kWh, while average access latency remains essentially unchanged. These results demonstrate that combining workload prediction, migration-cost awareness, and temporally varying carbon signals can improve the environmental efficiency of tiered storage management without compromising system performance.
References
[1] H. Han and M. Song, “QoE-aware video storage power management based on hot and cold data classification,” in Proceedings of the 28th ACM SIGMM Workshop on Network and Operating Systems Support for Digital Audio and Video, 2018, pp. 7–12.
[2] Y. Guan, Z. Ma, and L. Li, “HDFS optimization strategy based on hierarchical storage of hot and cold data,” Procedia CIRP, vol. 83, pp. 415–418, 2019.
[3] S. Ma, T. Zhao, J. Gu, and Y. Wang, “A hierarchical storage mechanism for hot and cold data based on temperature model,” in International Conference on Database and Expert Systems Applications. Springer, 2024, pp. 153–159.
[4] X. Xie, “MQHD: Dynamic load balancing method for energy-aware storage systems,” Scientific Programming, vol. 2020, no. 1, Art. no. 6843572, 2020.
[5] S. Zhizhuo, T. Yu-An, and L. Yuanzhang, “An energy- efficient storage for video surveillance,” Multimedia Tools and Applications, vol. 73, no. 1, pp. 151–167, 2014.
[6] M. Darwich, Y. Ismail, T. Darwich, and M. Bayoumi, “Cost-efficient storage for on-demand video streaming on cloud,” in 2020 IEEE 6th World Forum on Internet of Things (WF-IoT). IEEE, 2020, pp. 1–4.
[7] S. Raut, A. Saxena, W. Leggette, S. Patil, V. Hsu, and D. Archer, “Workload management for sustainability with storage systems in hybrid cloud deployment,” in 2023 IEEE Women in Technology Conference (WINTECHCON). IEEE, 2023, pp. 1–6.
[8] S. Köhler, B. Herzog, H. Herzog, M. Vögele, L. Wenzel, A. Polze, and T. Hönig, “Carbon-aware memory place- ment,” ACM SIGENERGY Energy Informatics Review, vol. 4, no. 3, pp. 39–45, 2024.
[9] M. Gebhard, “Carbon-aware systemd timers,” Science, 2024.
[10] V. Rao and A. A. Chien, “Understanding the operational carbon footprint of storage reliability and management,” ACM SIGENERGY Energy Informatics Review, vol. 4, no. 5, pp. 180–187, 2024.
[11] U. Gupta, Y. G. Kim, S. Lee, J. Tse, H.-H. S. Lee, G.-Y. Wei, D. Brooks, and C.-J. Wu, “Chasing carbon: The elusive environmental footprint of computing,” in 2021 IEEE International Symposium on High-Performance Computer Architecture (HPCA). IEEE, 2021, pp. 854–867.
[12] Y. Li and D. D. E. Long, “Which storage device is the greenest? Modeling the energy cost of I/O workloads,” in 22nd IEEE International Symposium on Modeling, Analysis, and Simulation of Computer and Telecommunication Systems (MASCOTS), 2014.
[13] E. Borba, E. Tavares, and P. Maciel, “A modeling approach for estimating performance and energy consumption of storage systems,” Journal of Computer and System Sciences, 2022.
[14] V. De Maio, R. Prodan, S. Benedict, and G. Kecskemeti, “Modelling energy consumption of network transfers and virtual machine migration,” Future Generation Computer Systems, vol. 56, pp. 388–406, 2016.
Downloads
Published
Data Availability Statement
The study uses the publicly available Azure Functions Blob Access Trace (2020). The processed data and the code used for the trace-driven simulations are available from the corresponding author upon reasonable request.
Issue
Section
License
Copyright (c) 2026 Nedasadat Taheri, Seyed Hossein Taheri, Houman Kosarirad (Author)
This work is licensed under a Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License.
This is an open access article distributed under the terms of the CC BY-NC-SA 4.0, which permits copying, redistributing, remixing, transformation, and building upon the material in any medium so long as the original work is properly cited.
