Home Power DIY

Grid Computing

Some consider this to be the “the third information technology wave” after the Internet and Web, and will be the backbone of the next generation of services and applications that are going to further the research and development of GIS and related areas.

Grid computing allows for the sharing of processing power, enabling the attainment of high performances in computing, management and services. Grid computing, (unlike the conventional supercomputer that does parallel computing by linking multiple processors over a system bus) uses a network of computers to execute a program. The problem of using multiple computers lies in the difficulty of dividing up the tasks among the computers, without having to reference portions of the code being executed on other CPUs.

Parallel processing

Parallel processing is the use of multiple CPU’s to execute different sections of a program together. Remote sensing and surveying equipment have been providing vast amounts of spatial information, and how to manage, process or dispose of this data have become major issues in the field of Geographic Information Science (GIS).

To solve these problems there has been much research into the area of parallel processing of GIS information. This involves the utilization of a single computer with multiple processors or multiple computers that are connected over a network working on the same task. There are many different types of distributed computing, two of the most common are clustering and grid processing.

The primary reasons for using parallel computing are:

Saves time.

Solve larger problems.

Provide concurrency (do multiple things at the same time).

Taking advantage of non-local resources – using available computing resources on a wide area network, or even the Internet when local computing resources are scarce.

Cost savings – using multiple “cheap” computing resources instead of paying for time on a supercomputer.

Overcoming memory constraints – single computers have very finite memory resources. For large problems, using the memories of multiple computers may overcome this obstacle.

Limits to serial computing – both physical and practical reasons pose significant constraints to simply building ever faster serial computers.

Limits to miniaturization – processor technology is allowing an increasing number of transistors to be placed on a chip.

However, even with molecular or atomic-level components, a limit will be reached on how small components can be.

Economic limitations – it is increasingly expensive to make a single processor faster. Using a larger number of moderately fast commodity processors to achieve the same (or better) performance is less expensive.

The future: during the past 10 years, the trends indicated by ever faster networks, distributed systems, and multi-processor computer architectures (even at the desktop level) clearly show that parallelism is the future of computing.

Distributed GIS

As the development of GIS sciences and technologies go further, increasingly amount of geospatial and non-spatial data are involved in GISs due to more diverse data sources and development of data collection technologies. GIS data tend to be geographically and logically distributed as well as GIS functions and services do. Spatial analysis and Geocomputation are getting more complex and computationally intensive. Sharing and collaboration among geographically dispersed users with various disciplines with various purposes are getting more necessary and common. A dynamic collaborative model -” Middleware” – is required for GIS application.

Computational Grid is introduced as a possible solution for the next generation of GIS. Basically, the Grid computing concept is intended to enable coordinate resource sharing and problem solving in dynamic, multi-organizational virtual organizations by linking computing resources with high-performance networks. Grid computing technology represents a new approach to collaborative computing and problem solving in data intensive and computationally intensive environment and has the chance to satisfy all the requirements of a distributed, high-performance and collaborative GIS. Some methodologies and Grid computing technologies as solutions of requirements and challenges are introduced to enable this distributed, parallel, and high-throughput, collaborative GIS application.

Security

Security issues in such a wide area distributed GIS is critical, which includes authentication and authorization using community policies as well as allowing local control of resource. Grid Security Infrastructure (GSI), combined with GridFTP protocol, makes sure that sharing and transfer of geospatial data and Geoprocessing are secure in the Computational Grid environment.

Conclusion

As the conclusion, Grid computing has the chance to lead GIS into a new “Grid-enabled GIS” age in terms of computing paradigm, resource sharing pattern and online collaboration.

Most industries today have become so dynamic that organisations have to consistently seek and adapt to change, in order to survive and prosper. Factors like more diversified customer preferences, technological advances, increased competitive threats and an intensified global economy are among the forces inducing change. Organisations need to become more adaptable embracing Charles Darwin’s view that “it is not the strongest of the species that survives, nor the most intelligent, but the one that is the most adaptable to change”.

A survey conducted by PricewaterhouseCoopers in March 2004 shows that 47% of the CEO’s of the US’s fastest growing companies believe that their most critical success factor is having flexible strategies to respond to accelerating business changes. However, many recently implemented Information Systems still tend to ignore this need for flexibility and at times are hard to scale and customise, thereby limiting the ability of an enterprise to react fast to its evolving business needs.

In the last two decades we constantly experienced a dramatic change in the way we store and process digital information. Every few years there has been an industry breakpoint; an important new computing concept that changed radically the way computers are used and Information Systems are implemented. Examples include graphical and more user-friendly interfaces, the clientserver concept and the Internet. Such factors have somehow aided and contributed to position computers as a necessary commodity. Additionally, with the constant drop in the cost of hardware, and better and cheaper network bandwidth, computers have become even more ubiquitous. The Internet has evolved tremendously and is today considered as probably the most effective communication medium. Whilst technology tends to evolve in a non-linear fashion, Moore’s Law has ensured that processing power has been increasing exponentially.

Though this is contributing to easier hoarding and dissemination of information, ICT professionals today still face tough challenges. ICT budgets grew rapidly in the late 90’s in anticipation of the Y2K problem. In these last years many ICT departments have been even asked to cut their budgets while they were expected to continue providing an appropriate information infrastructure so as to enable the organisations to augment their products and possibly gain a competitive edge. Hardware replacement cycles are perceived to have increased. Generally speaking, ICT budgets did not grow in these last years in line with the computational needs of the organisations; whilst workloads are still increasing, the capacities to handle them are not.

In some cases increasing a firm’s computational needs might end up in a lot of computational power which is not appropriately utilised. Why? Consider for example the utilization of a server machine. Most of the time its real processing capacity is not used at all. However maybe sometimes because a large and long process is executed or the number of connected users temporarily increase, the server might endup experiencing a processing overload. It has been estimated that on average a desktop computer uses only about 5% to 8% of its processing power (EuropeanCeo, 2005). Whilst, as Hendry (2004) reports, load balancing can aid in the distribution of processing and communication activity, similar servers that experience spikes in processor usage are barely used for the rest of the day and eventually end up with a large amount of unused computing capacity.

So the inevitable questions are, is it really feasible to increase and upgrade the firm’s single source of computational power if most of the time the existing processing power is not being used? How can we ensure that a firm’s computational resources are well balanced and allocated, so as to minimise wastage and eventually, justify any further investment in the ICT infrastructure?

The basic concept that gives insight to the answer to these questions extends back to the 70’s when the notion of distributed computing was born. Today, we are seeing increasing interest among business communities in what is termed as, Grid Computing.

Definition
World-renowned organisations are promoting the Grid in a big way and several definitions can be found. It has become a fashionable term. Dr. Ian Foster, a professor at the University of Chicago and director of the Distributed Systems Lab at Argonne National Laboratory, a pioneer in Grid Computing, provided his definition for the layman as being the “technology to enable the sharing of computing resources across institutional boundaries”. Research firm, Gartner, Inc., defines grid computing as a way to solve computing tasks using resources that are shared by more than one owner and coordinated to solve more than one problem.

The concept of Grid Computing was initially popular among academics, research and scientific communities. It was used for functions that required a substantial amount of computing power. However in these last years, an increasing number of organisations are early adopting and trying to reap benefits from this technology.

There are numerous examples where Grid computing has been applied. Among the research communities, Oxford University is using Grid technology to analyse 3.5 billion molecules to work out their cancer fighting potential. Same is being done by Stanford University in order to analyse the role protein plays in keeping people healthy. The search for Extraterrestrial Intelligence (SETI) project is another example. Here, volunteers download and install a free program so as to process and analyse massive amounts of data in search of evidence of possible radio transmissions from extraterrestrial life. When tallying up all the processing power that these PC’s provide, it’s like having one big supercomputer. Grid technologies also played a major role in identifying the world’s largest known prime number. This was part of the Marsenne project where scientists identified the 43rd Marsenne Prime 230,402,457-1. – a figure that contains 9,152,052 digits.

Business Applicability
Within business communities, the Grid concept is far more popular among large corporations. Baum, the publishing editor for Oracle Corporation, states that these corporations are initially attracted by the amount of savings that the technology can provide. Mainstay Partners conducted an ROI study to evaluate the enterprise grid technology platforms currently in use at seven participating companies. It was concluded that the adaptation of grid technology yielded an average of 43 percent savings in hardware cost. Much of the savings were credited to the shift from a large symmetric multiprocessor server to a number of lower cost servers. With the use of Grid technology the latter setup delivered similar or at times even more computational power than the larger system, however with fewer costs. Baum’s report adds that the grids within these companies were being used for a variety of applications, including enterprise resource planning (ERP), decision support, customer relationship management (CRM), and supply chain management (SCM).

Still, companies that operate in the financial services industry, drug discoveries and weather modeling are initially more prone to benefit from Grid technologies, as they are involved in complex scientific and mathematical calculations and therefore require an added amount of computational power. So are companies that tend to process large amounts of data for their business intelligence activities. However, organisations are increasingly being enticed to adopt Grid technologies even for their transactional based systems, given that Grids may further facilitate storage space Issues.

Challenges faced by Grid Computing
IDC, the market intelligence and advisory services firm, are referring to Grid computing as the fifth generation of computing, after client-server and multi-tier (Table 1).

Yet, according to IDC, the technology still needs to be ‘normalised’ and has to overcome various challenges. IDC believes that these concerns, in some cases, are more perception than reality, and as organisations gain more experience with this distributed approach, their concerns will be laid to rest.

Additionally, a research conducted by the 451 Group shows that software licensing, security and bandwidth matters are among the things that can disturb grid rollouts.

Conclusion
Whilst Grid computing still needs to find broad acceptance in the commercial space, yet, market analysts state that the technology is here to stay. As Tom Hawk, the general manager of Grid computing for IBM says, “The Web is about sharing information. The grid is about sharing resources”.