Jurnal dasar Komputasi

LIMITATIONS OF GRID COMPUTING AS ANSWER

COMPUTING RESOURCES

Abstraction

        Grid computing, an architecture of high-performance computer systems that utilize existing grid computing technologies (some of which are: Globus Toolkit 4, Condor, PVM, MPI) as a component builders. With the creation of this computing grid computing infrastructure, the researchers hoped would need computing resources can be met and, in turn, can increase the competitive level.

Keywords: grid computing, distributed computing, PVM (Parallel Virtual Machine)

1.      Introduction

a.       Computing Resources needs the development of e-Science

              Currently, the researchers are already well aware of the importance of the role of computers in the development of science and technology. The computer allows the researchers to create a virtual laboratory in a computer to perform experiments that would be very expensive if done in a physical laboratory or even impossible. Some parties even has its own name to describe the activities of the development of science and technology is a computer-based e-Science as 0.

b.       The high cost of computing resources

To experiment with using a computer in the context of the development of e-Science in general takes over the resources of high-performance computing (or also known as high performance computing). In the past few decades, high-performance computing resources can only be met by a computer which is categorized as supercomputers (such as the Cray X-MP computer, CDC, ILLIAC-IV). Supercomputer is able to meet the needs of e-Science researchers, but because the price is expensive, only a few of

The researchers were able to have it / use it. In line with the development of computer technology, both in terms of hardware and software, the current high-performance computing resources no longer be met by the computers category supercomputer. Even with computer technology known as grid computing, a number of computers that are commonly used in offices can be combined to jointly perform experiments as previously practiced by computer supe.

c.       Grid Computing as a Solution

For researchers in countries with limited economic capabilities, the solutions provided by grid computing technology is an alternative that should be considered very seriously. Development of science and technology, including those carried out with the help of a computer (e-Science), does not have to stop just because of limited funds. Grid computing technologies enable researchers utilize computing resources that have been around as much as possible. By using this technology, the researchers can combine computers that are in places that are geographically separated into a unified computer system. Combined many computers as a whole is able to provide computing resources equal or even more with the computer category supercomputer. Furthermore, these computer systems can be used simultaneously by researchers who also comes from agencies different location. On the whole, not just the technology of grid computing allows researchers to implement e-Science to develop science and technology at a cost that is relatively "affordable", but also can take advantage of existing computing resources as efficiently as possible is shared by many researchers.

2.      Discussion

a.       The evolution of Grid Computing

Grid computing technology is a technology that has been developed in a long time. In evolution we see the development of similar technologies ranging from Condor 0, followed by PVM (Parallel Virtual Machine) 0 and MPI (Message Passing Interface) 0, up to the Globus Toolkit 0 Since the beginning, the researchers in the field of high performance computing has used two approaches 0, (1) supercomputer, built a computer with high-performance hardware technology, and (2) multicomputer, build a computer system with an interconnection network technology and software. The first approach generally produces a high-performance computer, but worth very expensive so it can only be owned by a handful of parties. The second approach results in a computer system whose performance varies according to the number of computers involved and the configuration of the software used.

Although the price of a high-performance computer systems are built with more affordable multicomputer approach compared to supercomputers, its use is still limited. The network-based computer systems are generally applied to the computers connected in a local network (LAN). One reason is that network security issues have not been handled properly. In addition, system support software that allows the computers to work as one unit generally has a complex configuration so that users must have their own expertise before it can take advantage of the computer system.

In line with the development of Internet technology and computer-related technologies such as data communication protocols, network security technology, distributed pemgrograman technologies, programming languages and technologies that are independent of the architecture of the computer-based high-performance computer systems network becomes easier to implement and use.

b.      Grid Computing & Solutions Offered

In recent years, a group of researchers in the field of high performance computing seriously focusing on the development of network-based computer system as described above by using a technology known as grid computing technology 0.

Grid computing technology is a way of merging the resources owned many computers connected in a network to form a unified computer system with computing resources on a scale approaching the amount of computing resources from computers that make it up. Further, some or all of the computing resources can be used by users according their individual needs. Naming the "grid" to borrow a term used here in the electricity 0, where power plants are connected to each other to jointly supply the electricity needs of its users. Each user only uses a portion of the electrical power generated by all the power plants.

In contrast to its predecessor technologies such as Condor, PVM, or MPI, grid computing technology is complemented by components that allow the use of pooled computing resources more optimally and safely. To see the components of grid computing technology, will be described briefly here Globus Toolkit system developed by researchers at Argonne National Laboratory, United States 0 system is one of the Globus Toolkit grid computing technologies that are popular and widely used by parties- those who want to integrate their computing resources scattered into a single unit.

Specifically, the system Globus Toolkit that will be discussed here is the Globus Toolkit system version 4 (GT4) 0, which is the latest version of the Globus Toolkit system. GT4 systems built using Web Services technologies 0 which has evolved into a standard in distributed software development. Web Services technology allows GT4 adopts the concept of service-oriented (service-oriented) who use the service, not the hardware, as the basic component of the building. On top of Web Services is the GT4 build major components of the system following grid computing.

c.       Infrastruktur Grid Computing

With the increasing need for researchers to computing resources to perform the e-Science as already mentioned upfront and development of grid computing technology then several countries have taken the initiative to implement a grid computing infrastructure at the national level. Some examples of these include: India 0, 0 of Singapore, and Japan 0.

A grid computing infrastructure will be able to reduce the investment cost than if each institution must hold its own computing devices. Furthermore, the grid computing system requires the use of computing resources together to foster the spirit of collaboration among the researchers. One thing that is very positive.

See the benefits that can be provided by the existence of a grid computing infrastructure at the national level in this paper proposed a draft RI-GRID, that grid computing infrastructure at the level of the Republic of Indonesia, which aims to utilize computing resources that are research institutions both now and in the future so that it can be used by researchers in these countries to develop science and technology.

d.      Arsitektur GRID COMPUTING

Figure 1 below shows the architectural design of the grid computing infrastructure. As shown in the picture, GRID COMPUTING constructed by combining the grid computing systems that are in research institutions (GRID-2, 3, 4) into a single unit. The configuration of hardware and software systems at each institutional level can be different, but with grid computing technologies operate as GT4 on the connecting nodes of each system, the whole system to form a unity grid computing infrastructure. With this configuration, if needed, the user in an institution can utilize computing resources that are outside the institution.

Configuration shown in Figure 1 above does not require each system at the institutional level to change the configuration of each system significantly. If an institution has implemented a grid computing technology such as SUN Grid Engine (SGE) or network-based computing technologies such as PVM, MPI, Condor GT4 then the system can be configured to coordinate with each of these technologies.

 One of the prerequisites of the formation of GRID COMPUTING is the availability of a large-capacity backbone network to connect nodes liaison at each institution (must have a bandwidth ranging 2 Mbps to 155 Mbps).

1.      GRAM: Grid Resource Allocation and Management

This component is responsible for managing all the computing resources available in a grid computing system. This management includes the execution of the program on all computers belonging to the grid computing system, ranging from the initiation, monitoring, until scheduling (scheduling) and inter-process coordination.

             An interesting thing with GT4 system is its ability to work with the systems management of computing resources that have been there before, such as Condor, PVM, or MPI. With this mechanism, the programs that have been built previously not need to be rebuilt or if they should be modified, the minimum modification, if it will run in the GT4-based grid computing environment.

2.      RFT / GridFTP: Reliable File Transfer / Grid File Transfer Protocol

                  This component allows users to access data from the large computing nodes belonging to the grid computing system in an efficient and reliable. This is important because the performance of computing not only depend on how fast computers are incorporated in the grid computing system executing the program, but also how quickly the data required in the computing accessible. Keep in mind that, the data required by a process not always be on the computer that executes the process.

3.      MDS: Monitoring & Discovery Service

             This component allows the user to monitor system GT4 computational ongoing process so that problems can arise immediately known. Meanwhile, the discovery aspect of this component allows the user identi-fication of the existence of the following characteristics of computing resources.

4.      GSI: Grid Security Infrastructure

This component is responsible for the security of the grid computing system as a whole. This component also, which is one distinguishing feature of GT4 technology with its predecessor technologies such as PVM or MPI. With the implementation of security mechanisms integrated with the components of other grid computing, grid computing technology based systems such as GT4 accessible to the public (WAN) without lowering the level of safety.

GT4 security system built on standard components of a security that have been tested, which includes data protection, authentication, delegation, and authorization. The basic configuration GT4 assuming both users and services using security standards using standard X.509 public key.

3.      Clossing

High-performance computing systems based on grid computing technology is not synonymous with high-priced computer system. Furthermore, the grid computing infrastructure can be constructed by combining the resources of existing computing into a single unit which can then contribute to the development of science and technology. In fact, prinsipkolaborasi underlying grid computing technology can be a valuable lesson for us to apply it in the context of the lives of others.