GPU-aware resource management in heterogeneous cloud data centers

Kulkarni, Ashwin; Annappa, B.

doi:10.1007/s11227-021-03779-4

Cited by 5 publications

(2 citation statements)

References 17 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In [28] presented a multi-objective resource allocation approach based on VM shape and physical PM system, optimizing energy consumption using Particle Swarm Optimization (PSO). Their proposed VMPSO mechanism was validated in terms of violations and performance of VM migration mechanisms.…”

Section: Introductionmentioning

confidence: 99%

Enhanced Neural Network Metaheuristic Approach for Optimizing Virtual Machine Deployment in Cloud Data Centers

Salehi,

Fartash,

Torkestani

2024

Preprint

View full text Add to dashboard Cite

In recent years, the optimization of virtual machine placement (VMP) in cloud data centers has emerged as a crucial and complex challenge, garnering considerable attention from researchers. Due to its NP-Hard nature, metaheuristic algorithms have become a popular approach in addressing this problem. These algorithms typically initiate with the random generation of initial solutions, which are then iteratively refined using specific operators to find the optimal solution. Among these algorithms, the Neural Network Algorithm (NNA) stands out as a promising swarm optimization technique known for its strong global search capability, making it well-suited for tackling various intricate optimization tasks. However, NNA often suffers from slow convergence due to its limited exploitation capability, thereby hindering its practical utility in optimization problem-solving. To address this limitation, this study proposes an approach to enhance the convergence speed of NNA while achieving a better balance between exploitation and exploration. Specifically, we adopt a heuristic algorithm to generate initial solutions in a more suitable and less random manner for the VMP problem. Additionally, the search operators of the algorithm are optimized using Cauchy and Lévy distributions. We evaluate the performance of the Improved Biogeography-based Neural Network Algorithm (IBNNA) on synthetic datasets of varying dimensions and compare the results with existing similar algorithms in the literature. The experimental findings demonstrate that the proposed algorithm outperforms other methods in terms of physical machine utilization, convergence rate, and power consumption across most experimental scenarios.

show abstract

Section: Introductionmentioning

confidence: 99%

Enhanced Neural Network Metaheuristic Approach for Optimizing Virtual Machine Deployment in Cloud Data Centers

Salehi,

Fartash,

Torkestani

2024

Preprint

View full text Add to dashboard Cite

show abstract

“…Graphics Processing Unit (GPU) clusters offer notable advantages in efficiently managing parallel processing and computationally intensive tasks [1,23,41]. In recent years, owing to rapid advancements in deep learning (DL) frameworks [30,40,41], the unprecedented escalation in data volumes and the diversification of user demands [41], cloud service providers introduce various types of GPU computing cards to meet the requirements of the modern computational landscape [23,36].…”

Section: Introductionmentioning

confidence: 99%

Forecasting GPU Type Allocation via Demand Feature Extraction in Heterogeneous Clusters

Wang,

Shi,

Chen

et al. 2024

Preprint

View full text Add to dashboard Cite

The availability of multiple types of GPUs enhances parallel processing capabilities and provides users with a broader spectrum of options in heterogeneous GPU clusters. However, the inherent randomness and uncertainty in GPU type requests introduce complexity into resource allocation, resulting in disparities between requested and allocated GPU types and exacerbating imbalances in the GPU resource distribution. In this study, we perform a comprehensive analysis of the attributes associated with diverse GPU requests and allocations in heterogeneous clusters. By introducing the adaptive selection model (ASM), our approach not only predicts the demand for various GPU types but also extracts critical features from the resource demand. These features play a pivotal role in predicting GPU resource allocation, effectively bridging the gap between request and allocation. Moreover, we introduce spatial and temporal long short-term memory (ST-LSTM), a novel LSTM-based model that accommodates both spatial and temporal correlations in GPU type allocation. Our experimental efforts confirm the effectiveness of our proposed methodology through the utilization of real-world trace data from Alibaba cloud data centres. The ASM achieves an impressive accuracy rate of 87% and illuminates the key factors influencing user GPU type selection. Additionally, the ST-LSTM model demonstrates excellent performance with average RMSE and MAE values of 1.84 and 1.09, respectively.

show abstract

gVMP: A multi-objective joint VM and vGPU placement heuristic for API remoting-based GPU virtualization and disaggregation in cloud data centers

Siavashi

Momtazpour

2023

Journal of Parallel and Distributed Computing

View full text Add to dashboard Cite

GPU-aware resource management in heterogeneous cloud data centers

Cited by 5 publications

References 17 publications

Enhanced Neural Network Metaheuristic Approach for Optimizing Virtual Machine Deployment in Cloud Data Centers

Enhanced Neural Network Metaheuristic Approach for Optimizing Virtual Machine Deployment in Cloud Data Centers

Forecasting GPU Type Allocation via Demand Feature Extraction in Heterogeneous Clusters

gVMP: A multi-objective joint VM and vGPU placement heuristic for API remoting-based GPU virtualization and disaggregation in cloud data centers

Contact Info

Product

Resources

About