This paper describes the performance and interoperability issues that arise in the process of integrating cluster management systems into a wide-area networkcomputing environment, and provides solutions in the context of the Purdue University Network Computing Hubs (PUNCH). The described solution provides users with a single point of access to resources spread across administrative domains, and an intelligent translation process makes it possible for users to submit jobs to different types of cluster management systems in a transparent manner. The approach does not require any modifications to the cluster management software; however, call-back and caching capabilities that would improve performance and make such systems more interoperable with wide-area computing systems are discussed. 1. Introduction Cluster management systems manage access to workstations, servers, and specialized machines distributed across local-area networks. From a user's perspective, the systems provide a cent...

Online component markets can be costly for consumers to use, in terms of the time and e#ort spent understanding the components on o#er. This cost of understanding will deter consumers from reusing the available components. Software visualisations derived from the components' run-time behaviour can lessen the cost of understanding. We have developed a prototype tool called Spider for providing this functionality to producers and consumers. We discuss some of the issues involved, along with our experiences in implementing the prototype.

management architecture for computational grids. The design is based on two key realizations. One is that resource management involves a sequence of tasks that is best handled by a pipeline. As shown in the paper, this approach results in a scalable architecture for decentralized scheduling. The other realization is that static aggregation of resources for improved scheduling is inadequate in wide-area computing environments because the needs of users and jobs change with both, location and time. The described architecture addresses this problem by dynamically aggregating resources in a manner that continuously optimizes system response. This is accomplished by way of an active yellow pages directory that allows aggregation constraints to be (re)defined on the fly. An initial prototype of the active yellow pages service has been deployed in the PUNCH network computing environment. Experiences with the production PUNCH system and preliminary results from controlled experiments indicate that the active yellow pages service performs well.

. This paper outlines key issues that must be addressed in order to allow PVM-based programs to make effective use of resources within a wide area network-computing environment. Support mechanisms that allow unmodified PVM programs to be used within the PUNCH network-computing environment are also described. The mechanisms were found to be easy to implement, and preliminary experiences indicate that the described approach is well-suited for a networkcomputing environment. 1 Introduction Distributed applications are often built on top of message-passing standards such as PVM [1] and MPI [2]. These standards were originally designed for relatively structured environments, where users are aware of all available machines and have direct access to them. In this context, the emerging wide area networkcomputing environment presents two interesting challenges: 1) the large size of the environment makes it difficult for users to keep track of all available resources, and 2) the dynamic...

anner that is transparent to users. Users can run tools in the lab or at home. They can define and run simulations, and view the text and graphical output via their Web browsers. PUNCH consists of multiple, disciplinespecific "hubs." Of the hubs currently available on PUNCH, one is dedicated to computer architecture. It includes a growing number of independently developed tools for computer architecture research and education (for example, CacheSim5, DLX-View, Proteus, RSIM, Shade). Other hubs contain tools for parallel programming, computational electronics, and VLSI design. These hubs currently provide access to 50 tools from 13 universities and 6 vendors. They serve more than 800 users from within Purdue, across the US, and in Europe. Students in undergradu- Nirav H. Kapadia Renato J. Figueiredo Jos A.B. Fortes Purdue University THIS DISTRIBUTED NETWORK COMPUTER LETS USERS ACCESS AND RUN ENGINEERING TOOLS ANYTIME, ANYWHERE VIA<F

Governments increasingly need to collaborate with each other in order to solve global problems [1]. Information gathering, sharing, processing and retrieval are inherent to such collaborative efforts. Information systems that support international collaborations among governments face challenges that require research on how to • Protect, control access, filter, summarize, correlate and share information across agencies and organizations

This document describes PUNCH, a prototype network-computing infrastructure that supports the service-based computing paradigm outlined above. 2. PUNCH: The Purdue University Network Computing Hubs

terface to a wide-area network-computer? What are the types of operating system services that are required to allow remote collaborators to access and run legacy software applications via the Web? What are the implications of wide-area computing from a resource management perspective, and how can one reuse existing scheduling mechanisms? What does it take to manage and run a computing portal? How does one quantify the benefits of such a service, and what do the users think of it? Work on PUNCH has been funded by the National Science Foundation under grants DMR-9400415, MIPS-9500673, CDA-9617372, EEC-9700762, ECS-9809520, EIA-9872516, and EIA-9975275, by an academic reinvestment grant from Purdue University, and by a grant from the AT&T Foundation. Intel, SRC, and HP have provided equipment grants for PUNCH compute-servers. HPDC-9 Tutorial. Nirav H. Kapadia and Jose A. B. Fortes. 2 In addition to discussing the issues outlined above, the tutorial will b

anner that is transparent to users. Users can run tools in the lab or at home. They can define and run simulations, and view the text and graphical output via their Web browsers. PUNCH consists of multiple, disciplinespecific "hubs." Of the hubs currently available on PUNCH, one is dedicated to computer architecture. It includes a growing number of independently developed tools for computer architecture research and education (for example, CacheSim5, DLX-View, Proteus, RSIM, Shade). Other hubs contain tools for parallel programming, computational electronics, and VLSI design. These hubs currently provide access to 50 tools from 13 universities and 6 vendors. They serve more than 800 users from within Purdue, across the US, and in Europe. Students in undergradu- Nirav H. Kapadia Renato J. Figueiredo Jos A.B. Fortes Purdue University THIS DISTRIBUTED NETWORK COMPUTER LETS USERS ACCESS AND RUN ENGINEERING TOOLS ANYTIME, ANYWHERE V

Using tools aimed at promoting the reuse of existing components costs the user in the time and e#ort needed to install and understand the tool. These costs could counteract or subsume the benefits of reuse argued for by reuse practitioners, rendering the activity worthless. One approach to reducing these costs is to deploy the tools in an environment that the user is already familiar with, and has easy access to. We have chosen the web as just such an environment, and this choice can have a significant impact on the usability and utility of the tool. This paper discusses the di#culties that arise from our use of the web, and the manner in which we have partly overcome these di#culties.