Grid Service For The Prediction Of Computing Resource

The grid technology at first appeared in the 1990s, it might distribute using the Internet in different resources as well as the computer company and so on urban, different local, different national database, memory resources and software becomes the whole, to external looked, was similar to a complete supercomputer provides the user need the integrated each kind of information service, its core guiding principle was "the entire Internet is a computer". The grid, takes the establishment on Internet's one kind of new generation's infrastructure, in the domestic and foreign academic circle, as well as between the industrial world has aroused the widespread interest and enormously the echo. The grid technology's goal lies in through a random computer may provide the infinite computing resource and the computing power, may use for to turn on the voluminous information. The question which this kind of emerging technology may cause various enterprises could solve is unable to process before, the peak efficiency to use their original system, satisfied customer each kind of request, and reduced their computer resources the total cost.The emergence of grid technology to fully realize the purpose of resource sharing, it has many advantages, such as low-cost, efficient, easy to use and so on. Grid computing and emerging technologies not only increase the computing power, researchers working efficiency, and it can make idle system resources to be more fully utilized. At first glance, grid computing is also largely used in the academic and scientific research fields, but an increasing number of various large IT organizations, enterprises, research institutes and so want to take advantage of grid computing, the technology to accomplish their own number of projects. Moreover, the application of grid computing a number of developers have also seen this potential market, have stepped up research to early introduction of grid computing-related services.Now, outside the major IT companies have launched a number of forming grid computing products, while China was probably around the year 2000 began research in this area. At present, the foreign grid computing technology has developed very rapidly, the industry has launched a number of commercial grid computing platforms and related software packages, but the domestic related technology research is basically still in its early stages.In practical applications, how to make the user safe, effective and convenient use of resources is the need to focus on grid computing environment to consider. That is, when a user make job request, application resources, you need to use the computing resources can predict grid services on the current grid computing resources in the environment to predict, and thus be able to find out loads of free nodes available to the lower users.This paper relies on the National Natural Science Foundation project "Grid program execution performance of real-time monitoring and forecasting" Done. The project's main elements are:in different countries, different cities in the distribution of a large number of computer resources, in the calculation of the process of using the grid need to solve a number of large-scale computational problems, this time need to use resources (such as the host responsible for network bandwidth) continue to occur dynamically change. At this time, we need to find a computing grid of these ever-changing performance information, how can an accurate real-time access to this information, in order to effectively pool resources from the ever-changing select the appropriate computing resources is a serious problem. The project proposes to introduce real-time monitoring of resources, and through some of the historical monitoring data to train the prediction model, which through these models to predict the resource. We use this form can be effectively accomplish their own requirements.The application procedures for the implementation of the need to support real-time scheduling. In a support neither the number of appointments not support the overall priorities of shared environment to run the application, the application to ensure task completion time in order to increase the chance threshold must carefully choose which host to run the corresponding task. A real-time scheduling program is a middleware service that supports the application for the task of selecting the most suitable hosts. In addition to the task proposed a machine, the scheduler can also predict the task running time of the host. Modify the application to use this feedback task resource requests and run-time threshold until it found a host to meet the task's Run-time threshold. In the high-performance distributed computing systems, monitoring and management of the various components of distributed computing is very important. Large-scale distributed systems, such as computational grid, for different types of tasks require access to a large number of monitoring data, including performance analysis, adjustment, forecasting, scheduling, and error detection and diagnosis.Computational grid need for real-time, accurate dynamic performance information, even though the monitoring facilities to ensure the necessary work and be able to collect relevant information is very complex, but in high-performance distributed computing systems, monitoring and management of the various groups of distributed computing parts is very important. With distributed systems (such as computational grid) is constantly expanding in scale, complexity, increasing use of automated monitoring and management methods have become increasingly important.Grid Architecture Grid skeleton and soul of the grid, one of the core technology, so it is a grid a key component of the study. So far, the more important grid architecture there are three:one is the five-story hourglass structure; the other is the Open Grid Services Architecture (Open Grid Service Architecture, OGSA) and its initial base Facilities OGSI (Open Grid Service Infrastructure); The third is the WSRF (Web Services Resource Framework, Web Services Resource Framework), which is also based on the OGSA framework.