CSFPA: An intelligent hybrid workflow scheduling algorithm based upon global and local optimization approach in cloud

Abstract

Summary In cloud, the most prominent area is workflow scheduling due to its widespread application in different domains. It comes under the NP‐complete problem, henceforth researchers have suggested the nature‐inspired heuristics and metaheuristic algorithms but still, the results of these heuristics are not optimal. These algorithms are not competitive but complementary to each other, so in that case, hybridization may yield better results. Since then, researchers have started to combine different heuristics for better results. But in this process of hybridization, the two most important factors have not been considered yet, namely global and local optimization. The proposed work has considered the two most popular nature‐inspired workflow scheduling Cuckoo Search (CS) algorithm and Flower Pollination Algorithm (FPA) for hybridization. CS works best for global optimization, and FPA provides good results in the case of local optimization. The proposed work leveraged the benefits of both optimizations and proposed a novel hybrid algorithm entitled Cuckoo Search Flower Pollination Algorithm (CSFPA). The proposed hybrid algorithm uses multi‐objective functions to minimize cost and time and thus enhance the utilization of resources. The CSFPA algorithm has been compared with both single nature aspired algorithms and hybrid algorithm for performance evaluation.

A fine tune hyper parameter Gradient Boosting model for CPU utilization prediction in cloud

“…However, existing CPU resource prediction algorithms primarily focus on point value prediction, which suits short-term CPU load forecasting in centralized computer systems but falls short in accurately anticipating CPU utilization variations across large-scale distributed environments [3][4][5][6]. To meet this challenge, special prediction models designed for future CPU usage in cloud computing environments are needed.…”

A fine tune hyper parameter Gradient Boosting model for CPU utilization prediction in cloud environment

“…However, existing CPU resource prediction algorithms primarily focus on point value prediction, which suits shortterm CPU load forecasting in centralized computer systems but falls short in accurately anticipating CPU utilization variations across large-scale distributed environments [3][4][5][6]. To meet this challenge, special prediction models designed for future CPU usage in cloud computing environments are needed.…”