Grid Computings-1 1

7. Grid Computing Systems and Resource Management

7.1 Grid Architecture and Service Modeling The grid is a metacomputing infrastructure that brings together computers t

o form a large collection of compute, storage, and network resources to solve large-scale computation problems or to enable fast information retrieval by registered users or user group.

The coupling between hardware and software with special user applications is achieved by leasing the hardware, software, middleware, databases, instruments, and networks as computing utilities.

The goal of grid computing is to explore fast solutions for large-scale computing problems. Grid computing takes advantage of the existing computing resources scattered in a nation or internationally around the globe.

7.1.1 Grid History and Service Families

Grid Computings-1 2

The idea of the grid was pioneered by Ian Foster, Carl Kesselman and Steve Tuecke in 2001 paper [16].

The Globus Project supported by DARPA has promoted the maturity of grid technology with rich collection of software and middleware tools for grid computing.

Gids differ from conventional HPC clusters. Cluster nodes are more homogenous machines that are better coordinated to work collectiv

ely and cooperatively. The grid nodes are heterogeneous computers that are more loosely couples together over g

eographically dispersed sites.

Grid Computings-1 3

Four Grid Service Families Table 7.1 Four Grid Families Identified in the Great Global Grid (GGG)

Grid Service Protocol Stack Fig. 7.1 The layered grid service protocol and their relationship with the Internet

service protocols. Grid Resource

Table 7.2 summaries typical resources that are required to perform grid computing.

7.1.2 CPU Scavenging and Virtual Supercomputers CPU Scavenging and Virtual Supercomputers

The concept of creating a “grid” from unused resources in a network of computers is known as CPU scavenging.

The most famous simple is the SETI@Home, which applied over 3 million computers to achieve 23.37 Tflops as of Sept. 2001.

Grid Resource Aggregation Several assumptions

The compute nodes and other necessary resources for grids do not join or leave the system incidentally, except when some serious faults occur in the grid.

Grid Computings-1 4

Cloud resources are mostly provisioned from large data centers. The joining or leaving of some peers has little impact on the needed

functions of a P2P grid system. The availability and specification of these open resources is

provided by Grid Information Service (GIS) agencies. Virtual Organization

The grid is a distributed system integrated from shared resources to form a virtual organization (VO). The VO offers dynamic cooperation built over multiple physical organizations.

The virtual resources contributed by these real organizations are managed autonomously.

Figure 7.2

7.1.3 Open Grid Services Architecture (OGSA) The OGSA is an open source grid service standard jointly

developed by academia and the IT industry under coordination of a working group in the Global Grid Forum (GGF).

OGSA Framework The OGSA was built on two basic software technologies:

Grid Computings-1 5

The Globus Toolkit widely adopted as grid technology solution for scientific and technical computing.

Web services (WS 2.0) as a popular standards-based framework for business and network applications.

The OGSA is intended to support the creation, termination, management, and invocation of stateful, transient grid services via standard interfaces and conventions.

The OGSA framework specifies the physical environment, security, infrastructure profile, resource provisioning, virtual domains, and execution environment for various grid services and API access tools.

OGSA Interfaces Grid services demand special well-defined application interfaces. These interfaces provide resource discovery, dynamic service creation, lifetime

management, notification, and manageability. Grid Service Handle

The OGSA employs a “handle-resolution” mechanism for mapping a GSH to a GSR (Grid Service Reference).

Grid Service Migration Example 7.3 Grid Service Migration Using GSH and GSR

Grid Computings-1 6

7.1.4 Data-Intensive Grid Service Models Data Replication and Unified Namespace

This data access method is also known as caching, which is often applied to enhance data efficiency in a grid environment.

By replicating the same data blocks and scattering them in multiple regions of a grid, users can access the same data with locality of references.

Grid Data Access Models Four access models for organizing a data grid

Monadic model: is a centralized data repository model. Hierarchical model: is suitable for building a large data grid which has only

one large data access directory. Federation model: is better suited for designing a data grid with multiple

sources of data supplies. Hybrid model: combines the best features of a hierarchical and mesh

models.

Parallel vs. Striped Data Transfer Parallel data transfer opens multiple data streams for passing

subdivided segments of a file simultaneously. In striped data transfer, a data object is partitioned into a number of

sections, and each section is placed in an individual site in a data grid.

Grid Computings-1 7

7.2 Grid Projects and Grid System Built

7.2.1 National Grids and International Projects Table 7.4 Some National Grid Projects in Five Countries Table 7.5 International Grid Projects

Grid Computings-1 8

7.2.2 NSF TeraGrid in the US TeraGrid is an open scientific discovery infrastructure combining leaders

hip-class resources at 11 partner sites to create an integrated, persistent computational resource in the USA.

TeraGrid resources are integrated through SOA.

Grid Computings-1 9

7.2.3 DataGrid in the European Union Fig. 7.8 Hierarchical data distribution of the European Data Grid (EDG).