In utility-driven cluster computing, cluster systems need to know the specific needs of different users so as to allocate resources according to their needs. They are also vital in supporting service-oriented Grid computing that harness resources distributed worldwide based on users' objectives. Market-based resource management systems make use of real-world market concepts and behavior to assign resources to users. This paper outlines a taxonomy that describes how market-based resource management systems can support utility-driven cluster computing. The taxonomy is used to survey existing market-based resource management systems to better understand how they can be utilized.

Data Grids have been adopted as the platform for scientific communities that need to share, access, transport, process and manage large data collections distributed worldwide. They combine high-end computing technologies with high-performance networking and wide-area storage management techniques. In this paper, we discuss the key concepts behind Data Grids and compare them with other data sharing and distribution paradigms such as content delivery networks, peer-to-peer networks and distributed databases.

In this paper we characterize resource usage in six data centers with approximately 1,000 servers. The results substantiate common knowledge that computing resources are typically under-utilized. Utility computing has been proposed as a way of increasing utilization and hence e#ciency. Using an o#-line integer programming model we bound the potential gain in e#ciency for several di#erent utility computing models. In a detailed study of a subset of the data center servers, we found that utility computing o#ered the potential to reduce the peak and mean number of CPUs needed by up to 53% and 79%, respectively.

Abstract. In this paper we propose a new infrastructure for efficient job scheduling on the Grid using multi-agent systems and a Service Level Agreement (SLA) negotiation protocol based on the Contract Net Protocol. The agent-based Grid scheduling system involves user agents, local scheduler agents, and super scheduler agents. User agents submit jobs to Grid compute resources. Local scheduler agents schedule jobs on compute resources. Super scheduler agents act as mediators between the local scheduler and the user agents to schedule the jobs at the global level of the Grid. The SLA negotiation protocol is a hierarchical bidding mechanism involving meta-SLA negotiation between the user agents and the super scheduler agents; and sub-SLA negotiation between the super scheduler agents and the local scheduler agents. In this protocol the agents exchange SLA-announcements, SLA-bids, and SLAawards to negotiate the schedule of jobs on Grid compute resources. In the presence of uncertainties a re-negotiation mechanism is proposed to renegotiate the SLAs in failure. 1

Grid computing systems that have been the focus of much research activities in recent years provide a virtual framework for controlled sharing of resources across institutional boundaries. Security is one major concern in any system that enables remote execution. Several techniques can be used for providing security in Grid systems including sandboxing, encryption, and other access control and authentication mechanisms. The additional overhead caused by these mechanisms may negate the performance advantages gained by Grid computing. Hence, we contend that it is essential for the scheduler to consider the security implications while performing resource allocations. In this paper, we present a trust model for Grid systems and show how the model can be used to incorporate the security implications into scheduling algorithms. Three scheduling heuristics that can be used in a Grid system are modified to incorporate the trust notion and simulations are performed to evaluate the performance.

Abstract. This paper discusses the concept of grid towards achieving a complete definition using main grid characteristics and uses found in literature. Ten definitions extracted from main literature sources have been studied allowing the extraction of grid characteristics while grid uses are defined in terms of the different types of application support provided by grids. A grid definition is proposed using these characteristics and uses. This definition may be very useful to determine the limits of the grid concept as well as to explore new application fields in grid computing. In this sense, the extracted characteristics are employed to determine the potential benefits a grid infrastructure may provide to Computer Supported Collaborative Learning applications. 1

Resource management constitutes an important infrastructural component of a computational grid environment. The aim of grid resource management is to efficiently schedule applications over the available resources provided by the supporting grid architecture. Such goals within the high performance community rely, in part, on accurate performance prediction capabilities. This paper introduces a resource management infrastructure for grid computing environments. The technique couples application performance prediction with a hierarchical multi-agent system. An initial system implementation utilises the performance prediction capabilities of the PACE toolkit to provide quantitative data regarding the performance of complex applications running on local grid resources. The validation results show that a high level of accuracy can be obtained, that cross-platform comparisons can be easily undertaken, and that the estimates can be evaluated rapidly. A hierarchy of homogeneous agents are used to provide a scalable and adaptable abstraction of the grid system architecture. An agent is a representative of a local grid resource and is considered to be both a service provider and a service requestor. Agents are organised into a hierarchy and cooperate to provide service advertisement and discovery. A performance monitor and advisor has been developed to optimise the performance of the agent system. A case study with corresponding experimental results are included to demonstrate the efficiency of the resource management and scheduling system. The main features of the system include: hard quality of service support using PACE performance prediction capabilities; agent-based dynamic resource advertisement and discovery capabilities; simulationbased quantitative grid performance analysis and useroriented scheduling of local grid resources.

Abstract not found

This paper reports the development of D-RAJE (Distributed Resource-Aware Java Environment), a Java-based middleware platform that makes it possible to model and to monitor resources in a distributed environment. With this middleware, any kind of hardware or software resource can be modelled using standard Java objects, and services allow to discover local as well as remote resources, and to observe the state of these resources either locally or remotely. D-RAJE is meant to ease the development of adaptive, security-oriented, or QoS-oriented Java applications, as well as the development of platforms capable of supporting such demanding applications.

In this paper, we present a scalable authorization service, based on the concept of fine-grained access control (FGAC), for large-scale Grid infrastructures that span multiple independent domains. FGAC enables participating resource owners to specify fine-grained policies concerning which user can access can their resources under which mode. We argue that such an authorization service must be integrated with the resource broker service to avoid scheduling requests onto resources which do not authorize the user request. For this reason, we develop a novel resource broker service that integrates access control with resource scheduling. In our system, both resource owners and users define their resource access and usage policies. The resource broker schedules a user request only within the set of resources whose policies match the user credentials (and vice-versa). Since this process of evaluating authorization policies of resources and user, in addition to checking the resource requirement, can be a potential bottleneck for a large scale Grid, we also analyze the problem of efficient evaluation of FGAC policies. In this context, we present a novel method for policy organization and compare its performance with other strategies. Preliminary results show that the proposed method can significantly enhance performance. 1