An official website of the United States government
Efficient resource scheduling in cloud computing requires accurate and interpretable forecasts of virtual machine resource demand. Production cloud workloads are often heterogeneous, bursty, and time-dependent, making resource allocation difficult under changing workload conditions. Over-provisioning wastes infrastructure resources, while under-provisioning may degrade performance and cause service level agreement violations. This thesis presents an explainable, leakage-safe, and regime-aware machine learning framework for predicting virtual machine resource demand to support high resource utilization in cloud environments. Using real-world workload traces from the Google Cluster Workload/Usage Trace dataset, this study first develops an explainable machine learning approach based on gradient-boosted regression models for predicting CPU demand from historical workload behavior. SHAP explainability is applied to interpret model predictions and identify the most influential features affecting resource demand. The work is then extended to a leakage-safe forecasting setting using Google Cluster-Usage Traces v3, where scheduling-time feature sets are constructed to avoid post-execution feature leakage and time-ordered evaluation protocols are used to reduce identity leakage from recurring workloads. Three scheduling-relevant targets are examined: mean CPU demand, normalized memory pressure, and p95 CPU demand as an indicator of burst risk. LightGBM is compared with strong history-based baselines, including LastSeen and exponential moving average, across time-ordered, gap-based, cold-start, and warm recurring workload regimes. The results show that simple history-based predictors often achieve the strongest performance for warm recurring workloads, while machine learning provides its clearest benefit in cold-start or sparse-history settings where prior workload observations are limited. To further explain these regime differences, this research introduces causal-regime maps using PCMCI+ to identify regime-specific dependency structures among request variables, scheduling metadata, historical statistics, and resource-demand outcomes. Strict directed causal parent maps are used for causal-regime interpretation, while broader PCMCI+-guided structural feature maps are evaluated through graph-guided LightGBM forecasting models. Overall, this research advances resource demand prediction from a purely accuracy-focused task toward an explainable, leakage-safe, and mechanism-aware framework for cloud scheduling. The findings support hybrid scheduling policies that combine machine learning for cold-start workloads with lightweight history-based forecasting for recurring workloads, while using causal-regime structure to improve interpretability and guide future resource management decisions.
more »
« less
This content will become publicly available on December 3, 2026
An Explainable ML-based Approach to Predicting Virtual Machine Resource Demand for High Resource Utilization in Clouds
Efficient resource scheduling in cloud computing is crucial for maintaining service quality while optimizing infrastructure costs. However, dynamic workloads and varying user demands make virtual machine (VM) placement and resource management highly complex. Efficient resource scheduling remains a significant challenge. Over-provisioning leads to wastage, while under-provisioning can cause performance degradation and service level agreement (SLA) violations. In this paper, we propose an explainable machine learning (ML)-based approach to predict VM resource demand to enhance placement decisions and achieve high resource utilization in clouds. We train an XGBoost regression model to predict CPU demand based on historical usage data. Also, we employ SHAP (SHapley Additive exPlanations) to interpret the model's predictions and uncover key decision-driving features. To evaluate our approach, we experimented with a real-world workload from the Google Cluster Workload Trace. Experimental results demonstrate improved prediction accuracy and model transparency, offering a practical and interpretable solution for intelligent VM scheduling in real-world cloud environments.
more »
« less
- Award ID(s):
- 2400459
- PAR ID:
- 10706585
- Publisher / Repository:
- Springer Nature
- Date Published:
- Subject(s) / Keyword(s):
- Cloud Computing VM Scheduling Resource Allocation Prediction Machine Learning
- Format(s):
- Medium: X
- Location:
- Las Vegas, NV, USA
- Sponsoring Org:
- National Science Foundation
More Like this
-
-
Predictive VM (Virtual Machine) auto-scaling is a promising technique to optimize cloud applications’ operating costs and performance. Understanding the job arrival rate is crucial for accurately predicting future changes in cloud workloads and proactively provisioning and de-provisioning VMs for hosting the applications. However, developing a model that accurately predicts cloud workload changes is extremely challenging due to the dynamic nature of cloud workloads. Long- Short-Term-Memory (LSTM) models have been developed for cloud workload prediction. Unfortunately, the state-of-the-art LSTM model leverages recurrences to predict, which naturally adds complexity and increases the inference overhead as input sequences grow longer. To develop a cloud workload prediction model with high accuracy and low inference overhead, this work presents a novel time-series forecasting model called WGAN-gp Transformer, inspired by the Transformer network and improved Wasserstein-GANs. The proposed method adopts a Transformer network as a generator and a multi-layer perceptron as a critic. The extensive evaluations with real-world workload traces show WGAN- gp Transformer achieves 5× faster inference time with up to 5.1% higher prediction accuracy against the state-of-the-art. We also apply WGAN-gp Transformer to auto-scaling mechanisms on Google cloud platforms, and the WGAN-gp Transformer-based auto-scaling mechanism outperforms the LSTM-based mechanism by significantly reducing VM over-provisioning and under-provisioning rates.more » « less
-
Serverless computing, or Function-as-a-Service (FaaS), enables a new way of building and scaling applications by allowing users to deploy fine-grained functions while providing fully-managed resource provisioning and auto-scaling. Custom FaaS container support is gaining traction as it enables better control over OSes, versioning, and tooling for modernizing FaaS applications. However, providing rapid container provisioning introduces non-trivial challenges for FaaS providers, since container provisioning is costly, and real-world FaaS workloads exhibit highly dynamic patterns. In this paper, we design FaaSNet, a highly-scalable middleware system for accelerating FaaS container provisioning. FaaSNet is driven by the workload and infrastructure requirements of the FaaS platform at one of the world's largest cloud providers, Alibaba Cloud Function Compute. FaaSNet enables scalable container provisioning via a lightweight, adaptive function tree (FT) structure. FaaSNet uses an I/O efficient, on-demand fetching mechanism to further reduce provisioning costs at scale. We implement and integrate FaaSNet in Alibaba Cloud Function Compute. Evaluation results show that FaaSNet: (1) finishes provisioning 2,500 function containers on 1,000 virtual machines in 8.3 seconds, (2) scales 13.4× and 16.3× faster than Alibaba Cloud's current FaaS platform and a state-of-the-art P2P container registry (Kraken), respectively, and (3) sustains a bursty workload using 75.2% less time than an optimized baseline.more » « less
-
The low cost and rapid provisioning capabilities have made the cloud a desirable platform to launch complex scientific applications. However, resource utilization optimization is a significant challenge for cloud service providers, since the earlier focus is provided on optimizing resources for the applications that run on the cloud, with a low emphasis being provided on optimizing resource utilization of the cloud computing internal processes. Code refactoring has been associated with improving the maintenance and understanding of software code. However, analyzing the impact of the refactoring source code of the cloud and studying its impact on cloud resource usage require further analysis. In this paper, we propose a framework called Unified Regression Modeling (URegM) which predicts the impact of code smell refactor- ing on cloud resource usage. We test our experiments in a real-life cloud environment using a complex scientific application as a workload. Results show that URegM is capable of accurately predicting resource consumption due to code smell refactoring. This will permit cloud service providers with advanced knowledge about the impact of refactoring code smells on resource consumption, thus allowing them to plan their resource provisioning and code refactoring more effectively.more » « less
-
The past decades witnessed the fast and wide deployment of Internet. The Internet has bred the ubiquitous computing environment that is spanning the cloud, edge, mobile devices, and IoT. Software running over such a ubiquitous computing environment environment is eating the world. A recently emerging trend of Internet-based software systems is “ resource adaptive ,” i.e., software systems should be robust and intelligent enough to the changes of heterogeneous resources, both physical and logical, provided by their running environment. To keep pace of such a trend, we argue that some considerations should be taken into account for the future operating system design and implementation. From the structural perspective, rather than the “monolithic OS” that manages the aggregated resources on the single machine, the OS should be dynamically composed over the distributed resources and flexibly adapt to the resource and environment changes. Meanwhile, the OS should leverage advanced machine/deep learning techniques to derive configurations and policies and automatically learn to tune itself and schedule resources. This article envisions our recent thinking of the new OS abstraction, namely, ServiceOS , for future resource-adaptive intelligent software systems. The idea of ServiceOS is inspired by the delivery model of “ Software-as-a-Service ” that is supported by the Service-Oriented Architecture (SOA). The key principle of ServiceOS is based on resource disaggregation, resource provisioning as a service, and learning-based resource scheduling and allocation. The major goal of this article is not providing an immediately deployable OS. Instead, we aim to summarize the challenges and potentially promising opportunities and try to provide some practical implications for researchers and practitioners.more » « less
