LSTM Based Container Scheduling In Kubernetes

Container scheduling is a critical task in cloud computing environments, where resource demands of containers need to be predicted and allocated efficiently. This paper proposes a novel approach for container scheduling using LSTM, which is a type of recurrent neural network that can effectively model multi-dimensional time series data and capture temporal patterns. In this approach, historical resource usage of containers is used as input to the LSTM model, which predicts their future resource demands. To deploy the trained model, we use Kubernetes, an open-source container orchestration platform that provides flexible resource management and scalability features. We demonstrate the effectiveness of our approach by conducting experiments using real-world container traces and evaluating the scheduling performance. The results show our approach performs more efficiently compared to currently available container scheduling algorithms in cloud computing in achieving high resource utilization and minimizing the number of unscheduled containers. Our study highlights the importance of choosing the right orchestration platform, Kubernetes which is a suitable platform for container scheduling using LSTM due to its scalability, flexibility, and fault-tolerance features. Our work provides a promising direction for improving container scheduling in cloud computing.