DPS-LSTM: A Dual Prediction Strategy Using LSTM For Process And Storage Load Balancing In Containerized Cloud Environments

Traditional load balancing techniques of containers in cloud computing are largely aimed at distributing compute workloads with no consideration for storage balance. The lack of consideration leads to resource hotspots, resource wastage and higher system latency and exposure. Thus, the proposed work herein presents DPS-LSTM (Dual Prediction Strategy based on Long Short-Term Memory) an adaptive prediction-driven container load balancing approach incorporating process and storage considerations for efficient resource deployment. Unlike traditional models, DPS-LSTM employs temporal usage behaviors and addresses newly used containers, using LSTM neural networks to make predictions about future demand in processing work as well as storage utilization. The two-pronged strategy facilitates anticipatory and equitable load balancing with effective decisions between containers based on previous trends and probable forecasts. Use of predictions at processing and storage levels ensures enhanced throughput, reduced latency and enhanced cloud security using efficient container orchestration. DPS-LSTM is an effective, secure, and responsive solution to cloud management issues in our present era. It plots prevailing usage patterns against optimal profiles of resource allocation. In scenarios involving workload exceeding capacity, the new strategy independently activates an autonomous negotiation-procurement method for acquiring containers. Unlike before, when it totally relied on internal resources pre-allocated for project implementation, it aptly negotiates, bids, and purchases more containers from reliable third-party suppliers based on workload and cost-effectiveness criteria. The strategy thus ensures efficient and smooth unloading of extra workload, as well as uninterrupted operations throughout processing and storage. The autonomous strategy effectively integrates workload prediction with autonomous procurement, thus enabling it to adapt and perform well even with fluctuating workload.