- Ubiquitous Computing
- Web Services
- The Semantic Web
- Spaces Computing
- Peer-to-Peer Computing
- Collaborative Computing
- Dependable Systems
- Security
- Languages
- Pervasive Computing
- Cluster Concepts
- Distributed Agents
- Distributed Algorithms
- Distributed Databases
- Distributed Filesystems
- Distributed Media
- Distributed Storage
- Grid Computing
- Massively Parallel Systems
- Middleware
- Mobile and Wireless Computing
- Network Protocols
- Operating Systems
- Real-Time and Embedded Systems
- Commentary
- Endnotes
This chapter is from the book
This chapter is from the book
This chapter is from the book
Grid Computing
The grid-computing fitscape aims to contribute to the development and advancement of technologies that enable standard, universal access to computing power and resources, all of which are used in a manner similar to electrical power. A computational grid is conceptually similar to an electric power grid. Grid computing envisions the coupling of geographically distributed resources to offer consistent, inexpensive, location-agnostic access to a wide variety of resources, thereby enabling the aggregation and sharing of such things as supercomputers, computer clusters, SANs, distributed databases, embedded networks, instruments, nodes on an arbitrary network, and, ultimately, people. Solutions for large-scale, CPU-bound, and data-intensive NDC applications are naturals for a grid-computing approach. Once specific to particular applications, grid computing for general-purpose commercial use was first made widely available by Sun Microsystems.
Sun Microsystems was the first large computer vendor to make grid computing available for general-purpose commercial use.* On July 24, 2000, Sun announced the acquisition of Gridware, Inc., a privately owned commercial vendor of advanced computing resource management software that originated in Regensburg, Germany. Gridware developed resource management software, which was used primarily in compute-intensive, technical computing environments, such as electronic design automation. Its products were deployed to optimize utilization of workstations, servers, and dedicated compute farms, an area of strategic interest to Sun.
Thanks to the acquisition, Sun released a general-purpose grid-computing production that allows any organization to reap the benefits of such an approach, for example, the following benefits:
-
Specialized agents on each machine to identify and deliver the compute resources as needed
-
A GUI and command-line interface for user job submission and control
-
A queuing system to manage priorities and to assign jobs to available resources
Sun Microsystems was the first large computer vendor to make grid computing available for general-purpose commercial use.* On July 24, 2000, Sun announced the acquisition of Gridware, Inc., a privately owned commercial vendor of advanced computing resource management software that originated in Regensburg, Germany. Gridware developed resource management software, which was used primarily in compute-intensive, technical computing environments, such as electronic design automation. Its products were deployed to optimize utilization of workstations, servers, and dedicated compute farms, an area of strategic interest to Sun.
Thanks to the acquisition, Sun released a general-purpose grid-computing production that allows any organization to reap the benefits of such an approach, for example, the following benefits:
-
Specialized agents on each machine to identify and deliver the compute resources as needed
-
A GUI and command-line interface for user job submission and control
-
A queuing system to manage priorities and to assign jobs to available resources
Clear resource utilization benefits can be achieved with products like Gridware. If an organization can better aggregate the compute power of existing servers and desktop nodes, a highly scalable clusterlike resource (which can include thousands of processors) is the result. Many organizations have made heavy investments in compute resources; many of those nodes remain idle much of the time. Estimates of electricity consumed by Internet-connected nodes in the United States range from 2 percent to 8 percent,[29] which may not seem substantial, but any waste of processor capability is a waste of resources. The grid computing fitscape of NDC will help address better resource utilization in the aggregate. Interestingly, the Gridware product from Sun also reflects the trend toward ephemeralization with respect to software product cost; the basic engine is free.