A ticketing company wants to sell tickets for the Rolling Stones farewell performance. The IT department manager is worried that the company's servers and software will be overwhelmed after starting online ticket sales. But in fact, the company has not added dozens of servers and storage systems. The IT staff just turned on the switch and connected the company's backbone network with a "grid". As a result, the company sold 900,000 tickets in three minutes, and no customer was turned away because of the insufficient processing capacity of the system.
The above scenario is not out of reach. Grid, as a new network that can bring huge IT resources such as processing and storage capacity, can cope with temporary use. Grid computing connects a large number of computers in different places through a * * * shared network, thus forming a virtual supercomputer, which combines the redundant processor capabilities of computers in various places and can provide huge processing power for research and other data centralized applications. With grid computing, organizations that cannot afford to buy multimillion-dollar supercomputers can also take advantage of its enormous computing power.
Utopia of computing?
Rob Batchelder of Gartner believes that the concept of grid has always been a "utopia" in the field of computing. Although it has great prospects in scientific and technological applications, its biggest drawback is the lack of obvious commercial applications. Since it appeared in Europe and America in 1990s, the grid has been mainly used to help scattered university researchers analyze the data of particle accelerators and giant telescopes. However, in the past two years, the concepts of grid and GlobusToolkit have been widely used in research and education, and dozens of large-scale global projects have adopted these technologies to challenge the massive computing problems in scientific computing.
At present, although grid technology is mainly controlled by academic institutions, enterprises are following suit. In fact, the main sponsors of the Global Grid Forum include Unilever, a company famous for selling soap and ice cream. Like many enterprises that are studying and evaluating grid technology, Unilever itself is still secretive about how to use this technology. And Johnson &; Pharmaceutical companies such as Johnson & Johnson and Merck, and manufacturing companies such as BMW and Boeing have all used the processing power and storage space of this technology for simulation experiments. For example, can drugs protect cells from viruses? Will the wings of the plane break in the storm?
Genetic research is a natural application of grid technology, and the investment needed in this field is difficult to be borne by one enterprise. Biotechnology companies can use grid technology to analyze genetic data. Doctors can use grid technology to make three-dimensional models of patients' organs as an auxiliary means to diagnose diseases; The grid can handle data streams from store cash records or financial markets. Aviation, insurance, transportation, national defense and other industries will also benefit from it. From this point of view, grid computing is not an unreachable utopia, and its broad commercial application prospects are just around the corner.
Fight for control
Grid computing is called "the third wave of information technology" after Internet and Web, and it is expected to provide the next generation of distributed applications and services, which will have a far-reaching impact on research and information system development. Major IT vendors have been fighting for the control of grid computing for a long time.
Sun Company recently released the beta version of "Grid Engine" Enterprise Edition 5.3, which makes it easier to connect computer grids within the enterprise and provides better management and resource allocation. Grid engine software provides an open source version. Since its release in 2000, * * * has been downloaded 654.38+0.2 million times, and * * has 654.38+0.654 38+0.8 million CPUs managed by this software. PeterJeffcock, technical product marketing manager of Sun Company, believes that grid computing has three obvious stages: cluster grid, campus grid and global grid. At present, the release of GridEngine Enterprise Edition 5.3 makes Sun take a step towards functional campus grid. Sun also supports industry organizations such as AVAKI and Globus together with competitors, and actively participates in the formulation of open standards for grid computing.
Microsoft's research department has also participated in various distributed computing research projects, including Farsite (a fault-tolerant remote file system) and Millenium, a distributed system. Hewlett-Packard also said that it will provide Coolbase software, so that users can enjoy various computing devices through the Internet. Compaq announced that it is developing a global grid computing solution plan to provide software, hardware and technical support for customers seeking grid computing systems. Therefore, Compaq formed an alliance with PlatformComputing in Canada, and made full use of this technology to run CompaqTru64UnixAlpha server system and CompaqProLiant server in Linux, providing users with a complete, integrated and open grid solution. Compaq also established an advanced research center for grid computing to continue the research of this technology. Japanese companies are also eager to try grid computing. NTT announced that it would start a six-month grid computing experiment in mid-2002, with the participation of Intel and SGI.
In August this year, IBM announced that it would invest $4 billion in the field of grid computing and build 40 data centers around the world, officially entering the field of grid computing. IBM was selected by the British government to take charge of the National Grid project, which will connect computers of eight universities with a budget of $25 million. IBM is currently working with the University of Pennsylvania to connect several hospitals and build a complex computing grid. Participating hospitals can quickly use telemedicine data and enjoy the analysis program. A few days ago, IBM also announced a project called North Carolina Bioinformatics Science Grid, which involves 60 enterprises, universities and biomedical research companies. This is the first power grid project in the world, mainly involving the private sector. At this time, it is only three months before IBM enters the field of grid computing. It seems that IBM is determined to become the "leader" of grid technology.
So, does the implementation of this project mark the beginning of grid computing into commercial applications?
Standards are the key to success.
Just as TCP/IP protocol is the core of the Internet, it is necessary to define standard protocols and services to build grid computing. At present, standardization organizations such as Global Grid Forum, Object Management Group (OMG) which studies model-driven architecture, W3C, Globus.org, etc., which are dedicated to the study of web services and semantic WWW, are just around the corner.
In July this year, standardization organizations such as OMG, W3C, Grid Forum, and people from academic and business circles participated in the "Software Service Grid Seminar" to speed up the formulation of GGG standards. Then, Globus, another open source grid standard organization, also gathered to study the infrastructure of high performance computing connected by WAN. Globus is currently working on developing standard grid architecture and other technologies.
So far, there is no formal standard for grid computing, but on the core technology, relevant institutions and enterprises have reached an agreement: the Globus Toolkit developed by Argonne National Laboratory and the School of Information Science (ISI) of the University of Southern California has become the de facto standard for grid computing. 12 computer and software manufacturers including Entropia, IBM, Microsoft, Compaq, Cray, SGI, Sun, Veridian, Fujitsu, Hitachi and NEC have announced that they will adopt Globus Toolkit. As an open architecture and open standard infrastructure, Globus Toolkit provides many basic services needed to build grid applications, such as security, resource discovery, resource management, data access and so on. At present, all major grid projects are based on the protocols and services provided by Globus Tookit.
In addition to standards, security and manageability, the lack of IT talents is also an urgent problem to be solved in grid computing, otherwise it will not become the business framework of enterprises. Problems that are often neglected in the internal system environment, such as security, authentication and reliability, must be solved in any distributed environment. Brent Sleeper, a partner of research and consulting firm StencilGroup, believes: "This requires high-level architectural skills, not the programming language listed on your resume." If the global grid is connected, then we can borrow each other's unused resources, and the grid will be more powerful and flexible. Although this is also the ultimate goal of the power grid, connecting the power grids together will also bring political problems. IBM's grid construction for universities or Unilever's internal grid construction is just a simple IT decision, but the risk of connecting private grids to form a grid with larger capacity is much greater. Are competing grid providers willing to sell each other's redundant resources when customers need them? In addition, grid applications often involve a lot of data and calculations, and it is necessary to share security resources among organizations, which is impossible for the current Internet and network infrastructure. It seems that there are still many problems to be solved to realize the commercial application of grid computing.
However, if we imagine what we can do with unprecedented computing power, we will all understand that the prospect of building a global grid is almost irresistible. Rick Stevens, a scientist at Argonne National Laboratory in the United States, pointed out: "Just like the initial ARPANET became the center of the Internet, let's regard Teragrid as the prototype of forming a global grid center!"
Commercial application of grid
Biomedicine: Grid can provide computing power for drug developers to study the morphology and movement of drugs and protein molecules.
Engineering: Boeing, Ford and BMW are all trying to use grid computing for complex simulation and design.
Data collection/analysis: Manufacturing, petroleum processing, cargo transportation and even retail enterprises have to maintain expensive equipment, which often leads to problems and bad results. Like wireless sensors, the grid can store and process all transactions.