Resource Utilization Analysis And Deployment Cost Of Deep Learning Model VGG16 And Resnet50 On AWS And GCP

Abstract

This research analyzes performance comparison and cost efficiency of deploying deep learning models VGG16 and ResNet50 on two leading cloud platforms, namely Amazon Web Services (AWS) and Google Cloud Platform (GCP). Using public datasets from Kaggle, this research involves data cleaning, splitting, and model deployment using a uniform instance configuration (2 vectors). using a uniform instance configuration (2 vCPUs, 4 GB RAM, 10 GB Disk). Performance monitoring was done through Amazon CloudWatch and Google Cloud Monitoring to record parameters such as CPU utilization, memory usage, and operational costs. Results show that AWS excels in resource performance performance, while GCP offers a more cost-effective solution in certain scenarios. This study provides data-driven guidance for organizations to choose the optimal cloud platform based on their specific needs. optimal cloud platform based on their specific needs.