AI-driven Dynamic Workload Balancing for Real-time Applications on Cloud Infrastructure
Madhusudhan Dasari Sreeramulu, Abdul Sajid Mohammed, Dinesh Kalla, Nagesh Boddapati, N. Yuvaraj
Abstract
The specification provides a new, resource-efficient method of Dynamic Workload Balancing in AI-driven Real-time Applications over Cloud Infrastructure. The real-time application keeps processing data at high speeds, and it is too difficult to get or make this type of arrangement using our traditional cloud setups as the system employs artificial intelligence methods responsively reallocate resources between virtual machines in accordion with demanded quality of services slabs in behalf-to-capacity ratio. It is scheduling jobs to resources efficiently and delivering results in a timely fashion, which is necessary for real-time applications like video processing or stock trading. It uses machine learning models based on historical data to predict the resources required for upcoming tasks. These predictions are subsequently utilized to smartly assign resources across VMs, factoring in network latencies and inter-dependence. It allows the system to modify queues in real-time according to changes in workload patterns, performing a dynamic load balancing.