multi-objective

MULTI-OBJECTIVE OPTIMIZATION FOR RESOURCE EFFICIENCY IN CLUSTERED ARCHITECTURES

• Author(s): Kalesha Khan Pattan
• Published In: International Journal of Innovative Research in Engineering & Multidisciplinary Physical Sciences (IJIRMPS)
• Publication Date: Aug 2021
• ISSN: ISSN 2147-6799
• DOI:
• Impact Factor: 9.88

The paper proposes a multi-objective optimization framework for efficient resource management in clustered architectures with heterogeneous workloads. It jointly optimizes throughput, latency, and energy consumption using adaptive evolutionary and heuristic techniques. Experimental results show around 30% improvement in throughput and 25–30% reduction in latency and energy usage across cluster sizes. The framework integrates with tools like Kubernetes and Spark, enabling scalable, energy-aware, and cost-effective orchestration without extra hardware.

• Multi-Objective Resource Optimization Framework: Designed a hybrid optimization framework that simultaneously balances throughput maximization, latency minimization, and energy efficiency, addressing the limitations of traditional single-objective cluster scheduling approaches.

• Pareto-Based Intelligent Decision Modeling: Introduced a Pareto-optimal optimization strategy that generates non-dominated resource allocation solutions, enabling flexible, context-aware trade-off selection for diverse workloads and business requirements.

• Adaptive Learning-Driven Resource Management: Integrated predictive analytics and reinforcement learning to enable continuous, real-time adaptation of compute, memory, and network resource allocation under dynamic and heterogeneous workloads.

• End-to-End Design, Implementation, and Validation: Led the complete design, simulation, and performance evaluation of an intelligent cluster management system, demonstrating consistent throughput gains (~30%) and significant reductions in latency and energy consumption without additional hardware investment.

• Sustained Performance Gains at Scale: Achieved up to ~30% improvement in system throughput with near-linear scalability across increasing cluster sizes, enabling high-performance operation without additional hardware investment.

• Energy-Aware and Cost-Efficient Operation: Reduced energy consumption and operational overhead by approximately 25–30% through intelligent, multi-objective resource allocation that avoids over-provisioning and idle resource wastage.

• Scalable Distributed Deployment: Enabled efficient scaling across increasing cluster sizes by dynamically adapting replica strategies to workload intensity and node behavior.

  Adaptive Resource Management for Dynamic Workloads: Enabled real-time adaptation to fluctuating workloads and heterogeneous node capabilities using predictive analytics and reinforcement learning, ensuring balanced utilization and low latency under varying conditions.

• Production-Ready Integration and Deployment: Designed the framework for modular integration with modern orchestration platforms such as Kubernetes and Spark, making it directly applicable to cloud-native, edge-computing, and enterprise cluster environments.

• Academic, Research, and Industry Value: Provides a comprehensive reference implementation—covering architecture design, optimization algorithms, simulations, and empirical evaluation—suitable for research, advanced coursework, and real-world intelligent resource orchestration systems.

MULTI-OBJECTIVE OPTIMIZATION FOR RESOURCE EFFICIENCY IN CLUSTERED ARCHITECTURE

• Kalesha Khan Pattan
• International Journal of Innovative Research in Engineering & Multidisciplinary Physical Sciences
• ISSN 2147-6799
• License
  This research is shared for a academic and research purposes. For commercial use, please contact the author.
  Resources
  https://www.ijisae.org/
  Author Contact
  LinkedIn: https://www.linkedin.com/**** | Email: [email protected]