A. S. GRIMSHAW, A. NGUYEN-TUONG, AND W. A. WULF, Campus-wide computing: Early results using legion at the University of Virginia, Tech. Report CS-95-19, University of Virginia, Mar. 1995.  Home/Search   Document Details and Download   Summary   Related Articles   Check

....cost of communicating initial data and results is negligible compared to the cost of each subcomputation. Thanks to these properties, such computations can efiqciently use idle processors linked through the Internet. There have been many attempts to use idle processor time for useful computation [4, 11, 15, 16, 19, 22]. Using the Internet and idle processing time for distributed computation introduces difiqculties not found in other parallel systems. The first problem is the Internet itself. Communication performance, in terms of end to end throughput, latency, and reliability, is relatively poor. The second ....

....open the problem up to a greater number of potential contributors. Similar goals and means were used by the DESCHALL project which cracked the code in the RSA DES Challenge[22] The majority of existing systems with similar goals impose significant system re quirements. For example, Legion [11, 12] requires at least two daemons to run on each client and uses a specialized file system. If a client fails, the centralized configuration database must be manually updated if the client cannot be restarted. Piranha [4, 3] is built on a tuple space based coordination language, Linda [9] Piranha ....

A. S. GRIMSHAW, A. NGUYEN-TUONG, AND W. A. WULF, Campus-wide computing: Early results using legion at the University of Virginia, Tech. Report CS-95-19, University of Virginia, Mar. 1995.

....Hence, a distributed environment is the most economical means of exploiting parallelism in this case. There have been many attempts to use idle processor power to useful computation. Majority of such system impose too heavy system requirements to be useful for this problem. For example, Legion [9, 10] requires at least two daemons to run on each client and uses a specialized file system. If a client dies, the centralized configuration database must be updated if the client cannot be restarted. Piranha [4, 3] is built on top of the tuple space based coordination language Linda [7] Piranha ....

A. S. Grimshaw, A. Nguyen-Tuong, and W. A. Wulf. Campus-wide computing: Early results using legion at the University of Virginia. Technical report, University of Virginia, 1995.

....and show some preliminary results. 1. INTRODUCTION A metacomputing system, or metasystem, can be defined as a collection of computer systems, accessible via a networked environment, that users can perceive as a single computer [2] Two well known metasystem projects are Globus [5] and Legion [7]. The difference between the concept of metasystem and the classic concept of distributed operating system is subtle. A metasystem is a more dynamic environment, comprised of independent computer systems, even with unrelated operating systems. Performance of applications running on a metasystem ....

Grimshaw, A. S.; Nguyen-Tuong, A; Lewis, M. J. and Hyett, M. 1997. "Campus-Wide Computing: Early Results Using Legion at the University of Virginia". The International Journal of Supercomputer Applications and High Performance Computing, 11(2):129-143.

....The aim of metasystem research projects is typically to build powerful and user friendly hardware software environments, by providing uniform and transparent access to the resources and applications available from computer networks. Two well known metasystem projects are Globus[Fos97] and Legion[Gri97]. The difference between the concept of metasystem and the classic concept of distributed operating system is subtle. A metasystem is a more dynamic environment, comprised of independent computer systems, even with unrelated operating systems. The performance of applications running on a ....

A. S. Grimshaw, A. Nguyen-Tuong, M. J. Lewis and M. Hyett. "Campus-Wide Computing: Early Results Using Legion at the University of Virginia". The International Journal of Supercomputer Applications and High Performance Computing, 11(2):129-143, Summer 1997.

....and PSEWare [1] a toolkit for building PSEs for symbolic computations. More generic infrastructure for building PSEs is also under development, ranging from fairly simple RPC based tools for controlling remote execution to more ambitious and sophisticated systems such as Globus [7] and Legion [12] for integrating geographically distributed computational and information resources. However, most of these PSEs lack the ability to build up scientific applications by connecting and plugging software components together. Component Based Software Engineering (CBSE) 2] is receiving increasing ....

A. S. Grimshaw. Campus-Wide Computing: Early Results Using Legion at the University of Virginia. Int. Journal of Supercomputing Applications, 11(2), 1997.

....24 focused on symbolic computations) have been developed. More generic infrastructure for building PSEs is also under development. This infrastructure ranges from fairly simple RPC based tools for controlling remote execution [2, 32] to more ambitious and sophisticated systems, such as Legion [20] and Globus [14] for integrating geographically distributed computing and information resources. The Virtual Distributed Computing Environment (VDCE) developed at Syracuse University [34] is broadly similar to the PSE software architecture described in Sections 2 and 3. However, components in the ....

A. S. Grimshaw, A. Nguyen-Tuong, M. J. Lewis, and M. Hyett, \Campus-Wide Computing: Early Results Using Legion at the University of Virginia," Int. J. Supercomputing Appl. Vol. 11, No. 2, pp. 129-143, Summer 1997.

....PSEs focused on symbolic computations) have been developed. More generic infrastructure for building PSEs is also under development. This infrastructure ranges from fairly simple RPC based tools for controlling remote execution [1, 24] to more ambitious and sophisticated systems, such as Legion [18] and Globus [15] for integrating geographically distributed computing and information resources. The Virtual Distributed Computing Environment (VDCE) developed at Syracuse University [26] is broadly similar to the PSE software architecture described in Sections 2 and 3. However, components in ....

A. S. Grimshaw, A. Nguyen-Tuong, M. J. Lewis, and M. Hyett, "Campus-Wide Computing: Early Results Using Legion at the University of Virginia," Int. J. Supercomputing Appl. Vol. 11, No. 2, pp. 129-143, Summer 1997.

No context found.

GRIM97B Grimshaw, A. S., Nguyen-Tuong, A., Lewis, M., Hyett, M., Campus-Wide Computing: Early Results Using Legion at the University of Virginia, Journal of Supercomputing Applications and High Performance Computing, Vol. 11, No. 2, pp. 129-143, Summer 1997.

....NX operating system facilities such as pspart and cubeinfo provide information about processor reservation for Intel multicomputers. In a workstation environment several systems have implemented reservation schemes that permit workstation owners to withdraw their machines from the shared set [35][52] Machines also become unavailable if the amount of available computational resources is too little to be useful. The availability of communication bandwidth is a more difficult problem. On multicomputers the amount of communication bandwidth is dependent on the size and location of the ....

A.S. Grimshaw, A. Nguyen-Tuong, and W.A. Wulf," Campus-Wide Computing: Early Results Using Legion at the University of Virginia, CS-95-19, March 1995. 148

....the process of deciding where an job and its constituent tasks will run. A schedule is the set of machines assigned to a job. This paper describes the issues and constraints inherent in wide area scheduling, and proposes the design of a wide area scheduler for the Legion parallel processing system [8]. A number of research projects have addressed infrastructure support for wide area computing. Infrastructure support includes remote execution, managing heterogeneity, shared file systems, etc. Two of these projects are Schooner [9] and Legion [8] Perhaps the most popular widearea system is the ....

....for the Legion parallel processing system [8] A number of research projects have addressed infrastructure support for wide area computing. Infrastructure support includes remote execution, managing heterogeneity, shared file systems, etc. Two of these projects are Schooner [9] and Legion [8]. Perhaps the most popular widearea system is the World Wide Web that provides support for data sharing, but is not geared to high performance computing. The most widely used local area distributed system is PVM [12] but it is limited to support for communication, heterogeneity, and remote ....

A.S. Grimshaw, A. Nguyen-Tuong, and W.A. Wulf," Campus-Wide Computing: Early Results Using Legion at the University of Virginia, CS-95-19, March 1995.

....The objective of the Legion project at the University of Virginia 2 is to construct such a machine a virtual computer capable of supporting distributed and parallel transnational applications within a single logical namespace. A small scale prototype, the campus wide virtual computer (CWVC) [11], is operational and is currently configured with over 80 workstations and an IBM SP 2. Already we are experiencing frequent host failures in the CWVC. On the scale of the envisioned nation wide virtual computer, host failures will be a fact of life. Therefore, fault tolerance both at the ....

A. S. Grimshaw, A. Nguyen-Tuong and W. A. Wulf, "Campus-Wide Computing: Early Results using Legion at the University of Virginia", Technical Report CS-95-19, Department of Computer Science, University of Virginia, 1995

....science problems; provide a shared high performance resource for university researchers; provide a given level of service (as measured by turn around time) at reduced cost; and act as a testbed for the nationwide Legion. The prototype consists of over sixty workstations and is now operational. In [18] we present the performance of two production applications that we have used to test the efficacy of our approach: complib, a biochemistry application that compares DNA and protein sequences, and ATPG, an electrical engineering application that generates test patterns for VLSI circuits. The ....

....of our approach: complib, a biochemistry application that compares DNA and protein sequences, and ATPG, an electrical engineering application that generates test patterns for VLSI circuits. The performance results are encouraging. Table 1 presents early performance results for complib taken from [18]. 3.2 The Nationwide Demonstration Project Now that the CWVC has successfully demonstrated the efficacy of our approach on a small scale, we will turn our attention to the nationwide Legion. The first step is to identify potential member organizations that would be interested in participating in ....

A. S. Grimshaw, D. Shiflet, and A. Nguyen-Tuong, "Campus-Wide Computing: Early Results Using Legion at the University of Virginia," submitted to Supercomputing `94.

....because of the scope of the project we cannot hope to present the details needed to satisfy the serious reader in just eight pages. A series of technical reports are available at http: www.cs.virginia.edu legion. The site includes technical reports on experiences with the prototype system [6], the security model [14] resource management [7] persistence, and a large number of slides that have been used for various presentations. The current implementation and all design documents and interfaces are also available at the web site. 2.1 Objectives From our Legion vision we have ....

A.S. Grimshaw, A. Nguyen-Tuong, and W.A. Wulf, "Campus-Wide Computing: Early Results Using Legion at the University of Virginia," University of Virginia Computer Science Technical Report CS-95-19, March 1995.

....science problems; provide a shared high performance resource for university researchers; provide a given level of service (as measured by turn around time) at reduced cost; and act as a testbed for the nationwide Legion. The prototype consists of over sixty workstations and is now operational. In [14] we present the performance of two production applications that we have used to test the efficacy of our approach: complib, a biochemistry application that compares DNA and protein sequences, and ATPG, an electrical engineering application that generates test patterns for VLSI circuits. The ....

A. S. Grimshaw, D. Shiflet, and A. Nguyen-Tuong, "Campus-Wide Computing: Early Results Using Legion at the University of Virginia," submitted to Supercomputing `94.

....an application increases so does the chance that the application will fail due to a host processor failure. At the University of Virginia, we have experienced first hand the problems caused by host failures in distributed systems while developing and using a prototype for the Legion project 2 [13][14] The objective of Legion is to construct the software environment to enable a nation wide or world wide virtual computer capable of supporting distributed and parallel applications. Our current prototype, which we call the Campus Wide Virtual Computer (CWVC) contains a mix of over 90 ....

A. S. Grimshaw, A. Nguyen-Tuong and W. A. Wulf, "Campus-Wide Computing: Early Results using Legion at the University of Virginia", Technical Report CS-95-19, Department of Computer Science, University of Virginia, 1995.

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC