ABSTRACTBACKGROUNDMOTIVATION FOR GRID COMPUTINGSWiM --AN INTEGRATION OF MOBILE WIRELESS COMPUTING AND GRID COMPUTINGSWiM-Globus ImplementationComponentsClient IPAnonymous TokenStart-Date/TimeEnd-Date/Time134.43.2.02345-5677-9895-938412/11/03:12:0012/21/03:12:00Figure 5. Sample Notification output from the Registration Service.The source code for the Registration Service can be found on the appendix. The files are RegServerFile.java, RegistrationProvider.java, Registration.wsdd, and Registration.gwsdl. The wsdd file describe the service. The gwsdl file is the interface of RegistratonProvider.java in XML format.FUTURE WORKS[17] I. Foster, C. Kesselman, J. Nick, S. Tuecke.The Physiology of the Grid: An Open Grid Services Architecture for Distributed Systems Integration. In Open Grid Service Infrastructure WG, Global Grid Forum, June 22, 2002.[18] . I. Foster, C. Kesselman, S. Tuecke.The Anatomy of the Grid: Enabling Scalable Virtual Organizations. In International J. Supercomputer Applications, 15(3), 2001.AppendixIndex.htmlRegistrationProvider.javaRegServerFile.javaRegistration.wsddRegistration.gwdlNatboxGUIClient.javaNatboxProvider.javaNatboxServer.wsddNatbox.gwsdlSWiM-GlobusSWiM-Globus :Secure Wireless Mobile (SWiM) GridComputingUsing Globus Toolkit 3.0{alvy, gmassoud}@cs.ucla.eduABSTRACTIn this project we implemented several components of SWiM using Globus 3.0. SWiM, a proposal by Jiejun Kong, is a new paradigm that integrates mobile wireless computing and grid computing together as a single problem of network resource management. In this report, we will first give a quick background on grid computing andthen introduce a major limitation in wireless computing today, which is the lack of simpleand secure cross domain mobility. By first modeling heterogeneous networks as grids, SWiM then addresses this problem by providing a set of common procedures for roamingwireless nodes to enter. By carefully heeding the “hourglass” structure of the internet today, SWiM does not modify the IP protocol stack, but uses Network Address Translation (NAT) to facilitate cross domain mobility. In these past few months we have implemented a set of components: 1) a registration server that coordinates with NAT boxes and allows roaming mobile clients to easily register with. 2) NAT boxes that configures accordingly as delegated by the registration server. Security components of SWiM will not be covered in this project.BACKGROUNDGrid computing focuses on large-scale resource sharing and innovative applications. Grid service has several advantages over web service. A web service provides an interface to persistent states of a single domain. It is subject to centralize control and it can be often categorized as the application layer. An example of a web service would a personal home page. On the other hand, a grid service provides interfaceto transient states of distributed activities. It is subject to de-centralized coordination, butnot subject to centralized control. Therefore, it is often categorized as a middleware between the application and the network layer. An example of grid computing is NASA’s Information Power Grid (IPG) which provides NASA scientists and engineers a substantial increase in their ability to solve problems that depend on large-scale and distributed resources. MOTIVATION FOR GRID COMPUTING The Internet Protocol suite follows an “hourglass” architecture, where IP protocol—the simple, well-defined network layer, serves as the bottleneck of the hourglass—provides uniform connection to diverse local domains. This design choice not only contributes to a scalable Internet, but also to another well-known design principle, also known as “the End-to-End Argument. Because of the prohibitively high cost of implementing new functions in the simple, well-defined, and widely-deployed IP protocol, the End-to-End Argument suggests that it is appealing to provide new services at end terminals rather than inside the network.Grid computing is an appealing answer to address the challenge of adding new services to a heterogeneous large-scale network. The Grid is the emerging computational and networking middleware infrastructure designed to provide uniform and secure accessto interconnected resources in an extremely heterogeneous wide area network. The Grid middleware is a layer of software that improves inter-operability between the network and the heterogeneous applications. If applications require new supports from lower layers, there is little need to change the simple network layer. Instead, new functions can be added into the evolving middleware layer, which is loaded on end hosts and interacts with each other to facilitate resource sharing.SWiM --AN INTEGRATION OF MOBILE WIRELESS COMPUTING AND GRID COMPUTINGMobile wireless computing has emerged as a complementary technology to Internet computing. Mobile wireless computing and Grid computing share many similar characteristics in their resource sharing requirements. First, both paradigms are applicable to large-scale networks, especially the Internet. Second, both paradigms require that network resource located in different local domain are not subject to centralized control, but should be coordinated for the purpose of smooth service provisioning. Third, both paradigms require using standard, open, general-purpose protocols and interfaces in coordinating. These common characteristics have motivated our work to integrate Mobile Wireless Computing with Grid Computing. Figure 1: Modeling wireless LANs as SWiM gridIn Fig. 1, some grids (1,3,11,12 in the figure) have higher security restrictions, while the others are open to public. These autonomous, distributed, heterogeneous, and secured SWiM-grids provide the following new supports to mobile nodes: (1) Inside each SWiM-grid, resource sharing is highly controlled. To maintain security confidence comparable toexisting grids in wired networks, SWiM enforces a wired-equivalent resource sharing policy in each of its wireless grid; (2) To enable coordinated inter-grid mobile resource sharing without changing the standard IPv4 protocol suite, SWiM chooses to utilize network address translation. Given the heterogeneity of grid computing, SWiM’s design choice maximizes resource sharing service availability when mobile nodes roam into resource limited grids; (3) Following design principles of grid computing, all network resources in SWiM are coordinated by grid
View Full Document
Unlocking...