Next-Generation Data Center Architectures: Exploring the Evolution of Infrastructure with Composable and Disaggregated Technologies
Abstract
The increasing complexity of modern workloads—driven by cloud computing, big data, artificial intelligence (AI), and the Internet of Things (IoT)—has necessitated a fundamental transformation in data center design. Traditional monolithic architectures are proving inadequate in meeting demands for scalability, agility, and efficiency. This paper explores next-generation data center architectures, focusing on composable infrastructure and disaggregated storage as key enablers of flexible, scalable, and energyefficient data environments. Composable infrastructure allows for on-demand allocation of compute, storage, and networking resources via software-defined mechanisms, reducing overprovisioning and enhancing operational flexibility. Disaggregated infrastructure physically separates resources, enabling independent scaling and optimized performance for diverse workloads. Drawing on theoretical frameworks, empirical studies, and real-world case analyses (e.g., Google’s Jupiter, Microsoft’s Project Olympus), this paper critically evaluates the performance, energy efficiency, and deployment implications of these technologies. The findings reveal that composable and disaggregated systems significantly enhance resource utilization, reduce data center power consumption, and improve workload adaptability. Challenges such as interoperability, implementation complexity, and industry standardization are also discussed. The paper concludes with recommendations for future-ready data center planning and identifies research avenues for continuous innovation.
Keywords
Composable Infrastructure, Disaggregated Storage, Data Center Architecture, Energy Efficiency, Software-Defined Infrastructure, Cloud Computing, Scalability, Edge Computin