Abstract:AbstractAmazon web services provide resources equipped with graphics processing unit (GPU) devices, which are commonly used for parallel processing applications. When there is less demand for cloud computing instances, the unused resources are provided at low-cost pricing as GPU spot instances. The pricing of the GPU spot instances dynamically changes over time based on the long-term demand and supply of cloud resources in the spot market. The main aim of this research is to predict upcoming GPU spot instance pricing using a time series prediction model. Our second aim is to study the unique characteristics of GPU spot instances. This research will help researchers to exploit GPU spot instances across distinct regions in a cost-efficient way. In the existing literature, researchers used machine learning approaches including artificial neural networks (ANN), long short-term memory (LSTM) networks, and random forecasts. They also used classical univariate time series approaches including autoregressive integrated moving average (ARIMA) and autoregressive moving average (ARMA) models. However, deep-learning price-reduction methods are complex, require high computational power, and they cannot provide accurate price forecasts due to the varying and volatile environment. As such, the use of time series prediction models is more suitable under these circumstances. But choosing a time series prediction model with different environments including variance and volatility is a challenging task, and none of the extant literature provides such an exploratory analysis. This research used the linear autoregressive (AR) model, ARIMA model, exponential smoothing (ETS) model, and generalized autoregressive conditional heteroskedasticity (GARCH) for predicting upcoming GPU spot instance pricing. This research provides a comparative exploration regarding the use of these prediction models across distinct regions in a cost-efficient way. The results comparisons between AR, ETS, ARIMA, and GARCH models show clearly that the GARCH model provides better results for dynamic based short-term and middle-term GPU instance price prediction. The AR model provides best results for variance based middle-term and long-term GPU instance price prediction. The ETS model provides better results for smoothness based long-term GPU spot price prediction. The ARIMA model gives best results for fluctuations based long-term GPU spot price prediction.

Price Forecasting for Spot Instances in Cloud Computing

Time-Series Analysis for Price Prediction of Opportunistic Cloud Computing Resources

An exploration to graphics processing unit spot price prediction

Self-Adaptive Cloud Pricing Strategies with Markov Prediction and Data Mining Method

On Pricing Strategy Analysis for Amazon Spot Cloud Service

Applying Machine Learning in Cloud Service Price Prediction: The Case of Amazon IaaS

Maximizing Cloud Revenue using Dynamic Pricing of Multiple Class Virtual Machines

Forecasting Price Trend of Bulk Commodities Leveraging Cross-domain Open Data Fusion.

Cloud Index Tracking: Enabling Predictable Costs in Cloud Spot Markets

A Hybrid Approach for Probabilistic Forecasting of Electricity Price.

Dynamic Pricing and Placing for Distributed Machine Learning Jobs: an Online Learning Approach.

Role of Prediction in Data-driven Pricing Schemes

A Cloud Computing-Based Intelligent Forecasting Method for Cross-Border E-Commerce Logistics Costs

Statistical analysis of Amazon EC2 cloud pricing models

Cloud Performance Trend Prediction by Moving Averages

An Integrated Day-Ahead Electricity Price Forecasting of Spot Market Based on Multi-Source Historical Information

It Can Be Cheaper: Using Price Prediction to Obtain Better Prices from Dynamic Pricing in Ride-on-demand Services

Using a Predator-Prey Model to Explain Variations of Cloud Spot Price

Intelligent Prediction Algorithm of Cross-Border E-Commerce Logistics Cost Based on Cloud Computing

Dynamic Price Prediction in Ride-on-demand Service with Multi-source Urban Data

Deep Learning-Based Hybrid Model for Forecasting Locational Marginal Prices