Grid Computing

by Randy Demidovich

Grid Computing is an evolving network application that can be expected to become an essential and necessary component of the new global knowledge economy of the 21st Century.

Grid computing is known by a variety of names, including Utility Computing, Pay-As-You-Go Computing, and Virtual Computing. Why all the names? The concept of Grid Computing has existed in one form or another for over twenty years, but as the Internet exploded in popularity during the 1990’s, a number of specific applications have emerged that require more attention to its development. Grid Computing is now a driving force behind broadband telecommunication networks and is seen as a potential solution to some of society’s more intractable problems.

What Problem is Grid Computing Designed to Solve?

Grid networks are used to compute extremely large data sets. While defining what large means in respect to data sets is continually evolving, there is an easy way to explain the basic need. Institutions that are currently working on cutting edge research in physics, engineering, and medicine are required to process huge amounts of data more quickly.

Mapping the human genome, developing a cure for cancer, and engineering a full-scale model of astronomical data are all examples of research that require the management of vast quantities of data. Prior to the advent of grid computing all of this data had to be loaded onto an individual workstation for manipulation. Considering such tasks now require terabytes of storage and processing power beyond the capabilities of a single PC, it has become illogical and economically infeasible to even try to manage this data in the context of a single desktop, or even within a local area network.

to individual workstations. These computers would analyze the problem and send the data back to a main database. Ohio University and affiliated universities participated in one such program called SETI@home, a computing network that scanned the universe for signs of extraterrestrial life. SETI stands for search for extra-terrestrial intelligence. This incarnation of grid computing, while powerful, has already proven to be insufficient. The SETI project relied on a program that determined when a networked workstation was not in use, much like a screen saver, and started and stopped the SETI@home program so as to not obstruct the workstation user.

How Does Grid Computing Work?

The newest incarnation of Grid Computing harnesses some of the technological leaps in networking and computing that have emerged since the turn of the century. Recent developments in Grid Computing now allow users to manipulate programs in real time while not actually computing from their own workstations. This is accomplished through the use of supercomputers. Workstation users operate their programs in the same fashion as they do today, but their data is processed differently. In modern business networks, for example, data is sent to a file server that computes or stores the necessary data and sends back to the desktop a portion of the raw information. This portion is then processed by the workstation. The workstation still has to have the processing power to compute all of the necessary operations.

For instance, if a workstation user were manipulating photographs stored on a file server with Adobe Photoshop, then the workstation would be communicating with the file server to retrieve some subset of the pictures, store those pictures, run Photoshop, compute all of the user’s decisions through keystrokes and mouse movements, compute what effect those movements made within the program, and send a visual output to the monitor.

Within the newest incarnation of Grid Computing, the workstation is only required to retrieve visual data and report the actions taken by the user. All of the processing and computing is done by the supercomputer. Instead of having Photoshop installed and running on the workstation computer, the actions of the workstation user are sent to the supercomputer that analyzes the information, computes what effect that data has within Photoshop, and sends back data in the form of pixels. [1] These pixels are shown on the workstation monitor in a way that is similar to the way Web pages are displayed. In this format, Photoshop is now being operated within a virtual environment. The users’ experience with Photoshop is unchanged, but their workstations’ computing power has exponentially increased because of the access to the virtual processing power of the supercomputer.

The opportunity for workstation users to operate programs virtually will shape the upcoming IT landscape with regard to infrastructure and end-user hardware. Imagine the result if a cell phone had access to a ten terahertz supercomputer. Grid technology has the ability to make that a virtual reality. The first diagram below shows how grid computing is done today. A number of powerful servers control all of the processing power of the entire grid. This technology is currently available. The second diagram shows how a more centralized version of Grid Computing would operate. Notice how individual workstations only have to have an Internet connection to the local area network (LAN) of the virtual supercomputer.

Figure 1. Franklin, Curtis Jr. “Grid-dy Determination” Network World Fusion. January 3, 2003, accessed via http://www.nwfusion.com/research/2003/0106grid.html

Figure 2. Base One International Corporation “Diagram of Virtual Supercomputer Grid,” accessed via http://www.boic.com/b1mgrid.htm

What is the Background of Grid Computing and Who is Promoting It?

Grid computing grew out of the work of research and development communities around the world. To date this technology has been purely open source and has been relatively free from market pressures. [2] An open source program is developed as a non-profit venture. The coding is available publicly. Open coding has enabled grid computing to grow. Since there were few competitive pressures, Grid Computing strategies could be tested in academic institutions and research facilities, and modifications made.

The history of Grid Computing deployment can be described in three phases. The First Phase was the academic phase in which increasingly massive computing was a necessity to progress in numerous scientific fields. The necessity became more evident with the development of data collection devices that could analyze extremely complex information such as the protein structure of a muscle cell. Some form of Grid Computing was needed to fully describe the protein structure of a muscle cell given the massive amount of information that was present. Grid computing is currently utilized in academic endeavors such as the development of new algorithms, mapping the human genome, and atomic and molecular visualization applications. The power of Grid Computing continues to develop in academic institutions. [3]

The Second Phase of Grid Computing deployment was the corporate phase. This phase is very active today. For the moment, the largest obstacle is building a commercial broadband infrastructure with the bandwidth to support such a demanding network. The technology is available to create such a network and the perceived return on investment is high enough that corporations are now finding it more feasible to switch to a Grid Computing system than to continually upgrade workstations.

The two corporations most prominently promoting this change are IBM and Sun Microsystems. They are the companies that will be selling the hardware. Each has also been instrumental in developing the software to enable Grid Computing. Surprisingly, they have continued to develop open-source software, the hope being that open-source programming will be somewhat resistant to the current standardization problems found within other sectors of the IT community, as with wireless networking. [4]

Some corporations are developing and marketing what is called middleware. Middleware is software specifically designed for Grid Computing. It is called middleware because it is the programming that acts as the go between the end-user and the computing power source. Corporations require middleware that is specifically designed for their set of tasks. While there are numerous open source developments in middleware, companies such as SGI are creating a trend toward proprietary applications custom built for specific applications. Market pressures for the means to process vast amounts of data ensures the continued implementation of Grid Computing in the corporate environment. [5]

Phase Three is deployment on behalf of consumers. This phase is developing slowly but has so many promising applications that it is beginning to see accelerated activity. The possibilities for consumer-based Grid Computing are staggering. The fact that there is development in middleware, end-user interfaces, and entirely new applications before there is an infrastructure in place is a clear signal that Grid Computing networks for consumers will not be far behind.

The main benefit of Grid Computing for consumers will be the removal of IT hardware from the home. Most computer users have no clue how their computers work. Grid Computing could remove many of the technological problems and much of the frustration encountered by today’s users. With the computing done outside of the home, technical problems can more easily be resolved by IT professionals at the location of the server. End users would only need to know how to operate their interfaces and interact with the middleware. This is where the name Pay-As-You-Go Computing was derived. Instead of buying all of the computing power and licenses to software, end-users would pay for applications, processing power, and storage as a service.

Economic and Regulatory Concerns

Building the infrastructure for Grid Computing is likely to produce a strong return on investment, hence the concept can be economically viable. The capabilities and services of a Grid Computing-based network far exceed the abilities of any affordable workstations. This creates strong incentives for corporations to upgrade. New opportunities for Applications Service Providers (ASPs) will ensure that Grid Computing will develop in the consumer arena. There is money to be made offering a subscription -based computing services to the public.

The removal of advanced computing hardware and software from end-user premises, and the emergence of ASPs providing computing and storage services online, will create real savings for consumers and make homes and small offices a much more attractive market. The economic prospects for Grid Computing must be seen as positive as a result of the growing need for more powerful computer processing capabilities in the context of rising expectations of academic, corporate and consumer users.

The regulatory concerns for Grid Computing are different in each phase. The need for regulation during Phase One was strictly for compatibility and a satisfactory result was achieved through the development of open-source programming. There is more of an emphasis on regulation in Phase Two due to security issues.

Corporations want to protect their data from competitors and hackers. While the government has not become involved in regulating grid computing, regulators seek to hinder the advancement of malicious programming and provide more severe penalties for hackers. The security issues are being resolved with open-source programming that allows users to report bugs and possible security weaknesses and with the use of biometric equipment. The advancement of biometric technology such as fingerprint and retinal scanners have helped to secure access and prevent identity theft. [6]

The largest regulatory concerns will grow from the widespread deployment of consumer- based Grid Computing. Security will hopefully be somewhat resolved once this time comes, but there will have to be some oversight of the new service-based computing businesses. What the parameters of this oversight will be is as yet unforeseen, but the main factor will be the conditions under which Grid Computing is introduced to the public.

A nationwide grid network could be developed publicly or privately. If corporations offer the service there are too many possibilities to pin down what the outcome would be. Would it be stable? Would it be affordable? Would it ensure privacy? Who would have access? What programs could operate virtually? There are still many issues that need to be resolved before consumer-oriented Grid Computing can become a reality.

Societal and Cultural Concerns

Social issues are matters of pressing concern for Grid Computing developers. The reason is that grid computing is extremely disruptive to the status quo. Grid technology will bankrupt a number of existing IT companies and entirely change how IT professionals do their jobs. For this reason, there is a degree of resistance that has to be addressed. Market pressures will be sufficient for Grid Computing to be adopted by the corporate world.

The resistance will come mainly from bringing grid computing to the consumer market. When consumers are given the means to interface Grid Computing networks, they will have an unprecedented amount of computing power and bandwidth. This raises concern over intellectual property and copyright infringement. While these concerns are present in our present incarnation of the Internet, Grid Computing will give unprecedented access to individuals that may choose not to follow the law. Even if security threats concerning such malicious programming as viruses and worms are brought under control, censoring an individual’s use of bandwidth may become a difficult task due to the sheer volume of data that will be passing through the network.

Even when adoption of Grid Computing will have a salutary effect on society, leading to scientific breakthroughs, there will be a number of ethical dilemmas. As humanity strives for the knowledge that can be obtained via additional computing power, humanity will still have to adapt to the knowledge acquired.

Developments greatly benefiting mankind, such as gene manipulation for curing disease, the creation of nanomachines, and finding new applications for quantum mechanics, raise ethical concerns currently. Genetic manipulation is being questioned on all fronts, whether from genetically modified foods to the cloning of human beings. Without careful preparation there could be quite a cultural backlash to Grid Computing. This, of course, like any other technology depends solely on how the society uses it.

Conclusion

Grid Computing is a technology that is bound to emerge in our society. Whether or not economic, cultural, or regulatory concerns allow for centralized computing to come to fruition, the currently available forms of Grid Computing can be expected to continue to expand. The outlook for Grid Computing entering the corporate markets is already strong and may soon be adopted on a wide scale. [7] The reason for this is pure necessity. For large conglomerates to stay competitive they will have to be able to process huge amounts of data. The easiest and most cost effective way of doing this is through Grid Computing. Extreme computing power to the home may be slower coming, but the opportunities presented by multimedia search engines linked to such a powerful networks will enable new types of on-demand services to the public.

Notes:

[1] Nicolaisen, Nancy “Grid Computing Application” Faulkner Information Services. April 6, 2004.

[2] Gentzsch, Wolfgang “Riding the Grid Wave” Computer World. December 6, 2004, p 30.

[3] Nichols, Bill “The Science Grid” Byte.com. January 10, 2005 Accessed via Academic Search Premier

[4] Vaas, Lisa “GlobusWorld Conference Reveals Growing Pains” eWeek. February 14, 2005, p 21

[5] Stewart, Walter “The Business of Grid” Webcast accessed via http://www.gridtoday.com/webcasts/index.html

[6] Vijayan, Jaikumar “Guarding the Grid” Computer World. November 29, 2004, p 32-33

[7] McMillan, Robert “Sun Flips Switch on Grid” Infoworld. February 7, 2005, p 17

Bibliography

Base One International Corporation “Diagram of Virtual Supercomputer Grid” accessed via http://www.boic.com/b1mgrid.htm

Coffee, Peter “Grids Can Be Great” eWeek January 3, 2005, p 46.
Dikaiakos, Marios “Introduction” Scientific Programming. December 2004, p 199.

Franklin, Curtis Jr. “Grid-dy Determination” Network World Fusion. January 3, 2003 accessed via http://www.nwfusion.com/research/2003/0106grid.html

Gentzsch, Wolfgang “Riding the Grid Wave” Computer World. December 6, 2004, p 30.

McMillan, Robert “Sun Flips Switch on Grid” Infoworld. February 7, 2005, p 17.

Nichols, Bill “The Science Grid” Byte.com. January 10, 2005 Accessed via Academic Search Premier

Nicolaisen, Nancy “Grid Computing Application” Faulkner Information Services. April 6, 2004.

Rupley, Sebastion “The Biggest Grid Yet” PC Magazine. January 2005, p 25.
Stewart, Walter “The Business of Grid” Webcast accessed via http://www.gridtoday.com/webcasts/index.html

Tuecke, Kesselman, and Foster “The Anatomy of the Grid: Enabling Scalable Virtual Organizations” International Journal of Supercomputer Applications. 2001.

Vaas, Lisa “GlobusWorld Conference Reveals Growing Pains” eWeek. February 14, 2005, p 21.

Vijayan, Jaikumar “Guarding the Grid” Computer World. November 29, 2004, p 32-33.