What the study found
The study found that the proposed DTCF framework for virtual machine consolidation reduced energy consumption and SLA violations in high-load scenarios compared with the state-of-the-art Fuzzy-GWO method.
Why the authors say this matters
The authors say VM consolidation is important for balancing energy use and service level agreements in cloud data centers, especially when workloads are non-steady-state and resources such as disk I/O are involved. The study suggests the framework is intended to improve system sustainability and operational stability under strict constraints.
What the researchers tested
The researchers introduced the Dynamic Threshold Control and Three-Dimensional Resource Coordination Optimization Framework (DTCF) for VM consolidation. It combines a Wavelet-TCN-LSTM model, which is a hybrid forecasting model using wavelets, temporal convolutional networks, and long short-term memory, with a 3D-PADT mechanism for predicting and adjusting overload thresholds, a DMRCIW policy for identifying and reallocating high-risk workloads, and a NAP-DRL placement algorithm based on deep reinforcement learning.
What worked and what didn't
On the Google Cluster Trace, DTCF outperformed Fuzzy-GWO in the reported high-load scenarios. The abstract reports a 23.2% reduction in energy consumption and a 43.5% reduction in SLA violations; it also says the framework better exploits heterogeneous hardware capabilities while respecting resource constraints.
What to keep in mind
The abstract describes results from experiments on the Google Cluster Trace, so the reported performance is limited to that setting. The available summary does not describe additional limitations or failure cases beyond the general challenges the authors identified.
Key points
- DTCF was designed for proactive virtual machine consolidation in cloud data centers.
- The abstract reports that it reduced energy consumption by 23.2% and SLA violations by 43.5% versus Fuzzy-GWO in high-load scenarios.
- The framework combines workload prediction, dynamic threshold control, multi-resource coordination, and deep reinforcement learning for placement.
- The evaluation used the Google Cluster Trace.
- The abstract says the approach addresses multi-dimensional resource constraints, including disk I/O.
Disclosure
- Research title:
- Proactive VM consolidation cuts energy use and SLA violations
- Authors:
- Guanghao Yang, Biying Zhang, Yanping Chen, Youbo Lyu
- Institutions:
- Harbin University of Commerce
- Publication date:
- 2026-03-03
- OpenAlex record:
- View
Get the weekly research newsletter
Stay current with peer-reviewed research without reading academic papers — one filtered digest, every Friday.