JAWAHARLAL NEHRU TECHNOLOGICAL UNIVERSITY HYDERABAD B.Tech IV HIGH PERFORMANCE COMPUTING Previous Question Paper (July/August 2022 - R18)

A Computational Grid focuses on providing massive processing power to solve computationally intensive problems, such as simulations or complex mathematical modeling. It aggregates distributed high-performance computing resources to execute tasks that require significant CPU cycles, often utilizing techniques like task farming and parallel processing.

A Data Grid is designed to manage, store, and provide access to massive amounts of distributed data. Its primary goal is to facilitate data-intensive applications, such as large-scale scientific experiments (e.g., High Energy Physics) or global databases. It emphasizes data replication, caching, metadata management, and efficient data movement across geographically dispersed storage systems rather than raw processing speed.

Grid architectures are typically structured as a layered model to manage the complexity of heterogeneous, distributed resources. The most widely accepted framework is the Grid Architecture Layered Model:

• Fabric Layer: Provides the actual physical resources, including computers, storage systems, networks, and sensors. It interfaces with local resource managers.
• Connectivity Layer: Defines core communication and authentication protocols required for secure, network-based transactions between grid resources.
• Resource Layer: Uses the connectivity layer to manage individual resources. It handles resource discovery, allocation, and monitoring through protocols like GRAM (Grid Resource Allocation Manager).
• Collective Layer: Coordinates multiple resources. This layer includes services for directory services, co-allocation, scheduling, and data replication.
• Application Layer: The top layer where end-user applications reside, utilizing the services provided by the lower layers to execute tasks or access data.

Additionally, grid architectures are often categorized by their deployment model, such as Intragrids (within a single organization), Extragrids (between partner organizations), and Intergrids (a global, interconnected network of grids).