Scalable dimensioning of resilient Lambda Grids

Grids consist of the aggregation of numerous dispersed computational, storage and network resources, able to satisfy even the most demanding computing jobs. Due to the data-intensive nature of Grid jobs, there is an increasing interest in Grids using optical transport networks as this technology allows for the timely delivery of large amounts of data. Such Grids are commonly referred to as Lambda Grids.

An important aspect of Grid deployment is the allocation and activation of installed network capacity, needed to transfer data and jobs to and from remote resources. However, the exact nature of a Grid’s network traffic depends on the way arriving workload is scheduled over the various Grid sites. As Grids possibly feature high numbers of resources, jobs and users, solving the combined Grid network dimensioning and workload scheduling problem requires the use of scalable mathematical methods such as Divisible Load Theory (DLT). Lambda Grids feature additional complexity such as wavelength granularity and continuity or conversion constraints must be enforced. Additionally, Grid resources cannot be expected to be available at all times. Therefore, the extra complexity of resilience against possible resource failures must be taken into account when modelling the combined Grid network dimensioning and workload scheduling problem, enforcing the need for scalable solution methods. In this work, we tackle the Lambda Grid combined dimensioning and workload scheduling problem and incorporate single-resource failure or unavailability scenarios. We use Divisible Load Theory to tackle the scalability problem and compare non-resilient lambda Grid dimensioning to the dimensions needed to survive single-resource failures. We distinguish three failure scenarios relevant to lambda Grid deployment: computational element, network link and optical cross-connect failure. Using regular network topologies, we derive analytical bounds on the dimensioning cost. To validate these bounds, we present comparisons for the resulting Grid dimensions assuming a 2-tier Grid operation as a function of varying wavelength granularity, fiber/wavelength cost models, traffic demand asymmetry and Grid scheduling strategy for a specific set of optical transport networks.     

Performance analysis of allocation policies for interGrid resource provisioning

Several Grids have been established and used for varying science applications during the last years. Most of these Grids, however, work in isolation and with different utilisation levels. Previous work has introduced an architecture and a mechanism to enable resource sharing amongst Grids. It has demonstrated that there can be benefits for a Grid to offload requests or provide spare resources to another Grid. In this work, we address the problem of resource provisioning to Grid applications in multiple-Grid environments. The provisioning is carried out based on availability information obtained from queueing-based resource management systems deployed at the provider sites which are the participants of the Grids. We evaluate the performance of different allocation policies. In contrast to existing work on load sharing across Grids, the policies described here take into account the local load of resource providers, imprecise availability information and the compensation of providers for the resources offered to the Grid. In addition, we evaluate these policies along with a mechanism that allows resource sharing amongst Grids. Experimental results obtained through simulation show that the mechanism and policies are effective in redirecting requests thus improving the applications’ average weighted response time.     

Grid Economics: A Selective Discussion of Two Research Problems

The last 5 years have seen considerable discussion of various types of Grids—compute Grids, storage Grids, and data Grids. Using the checklist given in Foster (, 2002) to define a Grid, two important problems that arise in the context of resource sharing in Grid computing environments are discussed. First, the well documented problem in compute Grid environments that arises from the inability of consumers to accurately estimate their resource requirements is presented. This results in incorrect scheduling of requests for Grid resources and social welfare loss. To address this problem, two research proposals are briefly described. The first approach argues for the design of decision support tools to help users with resource estimation while the second approach studies the design of resource allocation mechanisms that can work with stochastic specifications of resource requirements. This is in contrast to the traditional point estimates of resource required by extant mechanisms. Next, resource provisioning and pricing problems that arise in data storage and retrieval Grids are described. These Grids differ fundamentally from compute Grids but share some economic characteristics with P2P file sharing networks. Drawing on this connection, pricing mechanisms and resource provisioning research is briefly discussed.     

A software tool for semi-automatic gridification of resource-intensive Java bytecodes and its application to ray tracing and sequence alignment

Computational Grids deliver the necessary computational infrastructure to perform resource-intensive computations such as the ones that solve the problems scientists are facing today. Exploiting Computational Grids comes at the expense of explicitly adapting the ordinary software implementing scientific problems to take advantage of Grid resources, which unavoidably requires knowledge on Grid programming. The recent notion of “gridifying” ordinary applications, which is based on semi-automatically deriving a Grid-aware version from the compiled code of a sequential application, promises users to be relieved from the requirement of manual usage of Grid APIs within their source codes. In this paper, we describe a novel gridification tool that allows users to easily parallelize Java applications on Grids. Extensive experiments with two real-world applications - ray tracing and sequence alignment - suggest that our approach provides a convenient balance between ease of gridification and Grid resource exploitation compared to manually using Grid APIs for gridifying ordinary applications.     

Validating and Scaling the MicroGrid: A Scientific Instrument for Grid Dynamics

Large-scale Grids that aggregate and share resources over wide-area networks present major challenges in understanding dynamic application and resource behavior for performance, stability, and reliability. Accurate study of the dynamic behavior of applications, middleware, resources, and networks depends on coordinated and accurate modeling of all four of these elements simultaneously. We have designed and implemented a tool called the MicroGrid which enables accurate and comprehensive study of the dynamic interaction of applications, middleware, resource, and networks. The MicroGrid creates a virtual Grid environment – accurately modeling networks, resources, the information services (resource and network metadata) transparently. Thus, the MicroGrid enables users, Grid researchers, or Grid operators to study arbitrary collections of resources and networks. The MicroGrid includes the MaSSF online network simulator which provides packet-level accurate, but scalable network modeling. We present experimental results with applications which validate the implementation of the MicroGrid, showing that it not only runs real Grid applications and middleware, but that it accurately models both their and underlying resource and network behavior. We also study a range of techniques for scaling a critical part of the online network simulator to the simulation of large networks. These techniques employ a sophisticated graph partitioner, and a range of edge and node weighting schemes exploiting a range of static network and dynamic application information. The best of these, profile-driven placement, scales well to online simulation of large networks of 6,000 nodes using 24 simulation engine nodes.     

Modeling and Supporting Grid Scheduling

Grid resource management systems and schedulers are important components for building Grids. They are responsible for the selection and allocation of Grid resources to current and future applications. Thus, they are important building blocks for making Grids available to user communities. In this paper we briefly analyze the requirements of Grid resource management and provide a classification of schedulers. Then, we define an extensible formal model for Grid scheduling activities, and characterize the general Grid scheduling problem. Finally, we provide a reference architecture for the support of our model and discuss different aspects of architectural implementations.     

Using a Simple Prioritisation Mechanism to Effectively Interoperate Service and Opportunistic Grids in the EELA-2 e-Infrastructure

Grids currently in production can be broadly classified as either service Grids, composed of dedicated resources, or opportunistic Grids that harvest the computing power of non-dedicated resources when they are idle. While a service Grid provides high and well defined levels of quality of service, an opportunistic Grid provides only a best-effort service. Nevertheless, since opportunistic Grids do not require resources to be fully dedicated to the Grid, they have the potential to assemble a much larger number of resources. Moreover, these Grids cater very well to the execution of the so-called embarrassingly parallel applications, a type of application that is frequently found in practice, and that comprises the largest portion of the typical workload processed in production Grid systems. The EELA-2 e-infrastructure is comprised of a service Grid and an opportunistic Grid that federates computing resources from scientific institutions in both Europe and Latin America. Due to the complementary characteristics of these two types of Grids, a lot of attention has recently been placed in how to interoperate them. In this paper we focus on the less studied problem of assessing the feasibility of such interoperation. We analyse different prioritisation policies that define when the resources of one Grid can be used to run jobs originating from the other. Our results show that in the absence of a suitable prioritisation policy, the benefits that the users of one Grid may have, frequently come with an important negative impact on the users of the other Grid. We also show that a simple reciprocation mechanism is capable of arbitrating the interoperation in such a way that, whenever possible, users profit from the interoperation and, in no case, this benefit leads to a noticeable reduction on the quality of service that the users would experience were the Grids not to interoperate. We conclude discussing how we have implemented, in the context of the EELA-2 project, this prioritisation mechanism, allowing the effective interoperation of a service Grid based on the gLite middleware with an opportunistic Grid that uses the OurGrid middleware.     

计算网格中资源调度研发现状及展望

网格在解决分散式资源共享方面具有广阔的应用前景,使得它成为下一代互联网瞩目的焦点.而计算网格为资源的需求和供给提供定量分析,激励资源拥有者加入网格的同时保证资源消费者合理有效地使用资源.介绍计算网格通用模型GRACE的基础上,针对该模型中的热点问题--资源管理和调度展开讨论,分析比较当前的几种调度方案,为进一步提出高效完备的资源调度算法作准备.     

Towards supporting multiple virtual private computing environments on computational Grids

Grid computing now becomes a practical computing paradigm and solution for distributed systems and applications. Currently increasing resources are involved in Grid environments and a large number of applications are running on computational Grids. Unfortunately Grid computing technologies are still far away from reach of inexperienced application users, e.g., computational scientists and engineers. A software layer is required to provide an easy interface of Grids to end users.To meet this requirement HEAVEN (Hosting European Application Virtual ENvironment) upperware is proposed to build on top of Grid middleware. This paper presents HEAVEN philosophy of virtual computing for Grids – a combinational idea of simulation and emulation approaches. The concept of Virtual Private Computing Environment (VPCE) is thereafter proposed and defined. The design and current implementation of HEAVEN upperware are discussed in detail. Use case of Ag2D application justifies the philosophy of HEAVEN virtual computing methodology and the design/implementation of HEAVEN upperware.     

Reputation-based dependable scheduling of workflow applications in Peer-to-Peer Grids

Grids facilitate creation of wide-area collaborative environment for sharing computing or storage resources and various applications. Inter-connecting distributed Grid sites through peer-to-peer routing and information dissemination structure (also known as Peer-to-Peer Grids) is essential to avoid the problems of scheduling efficiency bottleneck and single point of failure in the centralized or hierarchical scheduling approaches. On the other hand, uncertainty and unreliability are facts in distributed infrastructures such as Peer-to-Peer Grids, which are triggered by multiple factors including scale, dynamism, failures, and incomplete global knowledge.In this paper, a reputation-based Grid workflow scheduling technique is proposed to counter the effect of inherent unreliability and temporal characteristics of computing resources in large scale, decentralized Peer-to-Peer Grid environments. The proposed approach builds upon structured peer-to-peer indexing and networking techniques to create a scalable wide-area overlay of Grid sites for supporting dependable scheduling of applications. The scheduling algorithm considers reliability of a Grid resource as a statistical property, which is globally computed in the decentralized Grid overlay based on dynamic feedbacks or reputation scores assigned by individual service consumers mediated via Grid resource brokers. The proposed algorithm dynamically adapts to changing resource conditions and offers significant performance gains as compared to traditional approaches in the event of unsuccessful job execution or resource failure. The results evaluated through an extensive trace driven simulation show that our scheduling technique can reduce the makespan up to 50% and successfully isolate the failure-prone resources from the system.     

Multi-Grid, Multi-User Workflows in the P-GRADE Grid Portal

Computational Grids connect resources and users in a complex way in order to deliver nontrivial qualities of services. According to the current trend various communities build their own Grids and due to the lack of generally accepted standards these Grids are usually not interoperable. As a result, large scale sharing of resources is prevented by the isolation of Grid systems. Similarly, people are isolated, because the collaborative work of Grid users is not supported by current environments. Each user accesses Grids as an individual person without having the possibility of organizing teams that could overcome the difficulties of application development and execution more easily. The paper describes a new workflow-oriented portal concept that solves both problems. It enables the interoperability of various Grids during the execution of workflow applications, and supports users to develop and run their Grid workflows in a collaborative way. The paper also introduces a classification model that can be used to identify workflow-oriented Grid portals based on two general features: Ability to access multiple Grids, and support for collaborative problem solving. Using the approach the different potential portal types are introduced, their unique features are discussed and the portals and Problem Solving Environments (PSE) of our days are classified. The P-GRADE Portal as a Globus-based implementation for the classification model is also presented. The work described in this paper is supported by the Hungarian Grid project (IHM 4671/1/2003), by the Hungarian OTKA project (No. T042459) and a collaboration project with the University of Reading.     

A Launch-time Scheduling Heuristics for Parallel Applications on Wide Area Grids

Large and dynamic computational Grids, generally known as wide-area Grids, are characterized by a large availability, heterogene- ity on computational resources, and high vari- ability on their status during the time. Such Grid infrastructures require appropriate schedule mechanisms in order to satisfy the application performance requirements (QoS). In this paper we propose a launch-time heuristics to schedule component-based parallel applications on such kind of Grid. The goal of the proposed heuristics is threefold: to meet the minimal task computation- al requirement, to maximize the throughput between communicating tasks, and to evaluate on-the-fly the resource availability to minimize the aging effect on the resources state. We evaluate the proposed heuristics by simulations applying it to a suite of task graphs and Grid platforms randomly generated. Moreover, a further test was conducted to schedule a real application on a real Grid. Experimental results shown that the proposed solution can be a viable one.     

A super-peer model for resource discovery services in large-scale Grids

As deployed Grids increase from tens to thousands of nodes, peer-to-peer (P2P) techniques and protocols can be used to implement scalable services and applications. The super-peer model is a novel approach that helps the convergence of P2P models and Grid environments and can be used to deploy a P2P information service in Grids. A super-peer serves a single physical organization in a Grid, and manages metadata associated to the resources provided by the nodes of that organization. Super-peers connect to each other to form a peer network at a higher level. This paper examines how the super-peer model can handle membership management and resource discovery services in a multi-organizational Grid. A simulation analysis evaluates the performance of a resource discovery protocol; simulation results can be used to tune protocol parameters in order to increase search efficiency.     

Designing a resource broker for heterogeneous grids

Grids provide uniform access to aggregations of heterogeneous resources and services such as computers, networks and storage owned by multiple organizations. However, such a dynamic environment poses many challenges for application composition and deployment. In this paper, we present the design of the Gridbus Grid resource broker that allows users to create applications and specify different objectives through different interfaces without having to deal with the complexity of Grid infrastructure. We present the unique requirements that motivated our design and discuss how these provide flexibility in extending the functionality of the broker to support different low‐level middlewares and user interfaces. We evaluate the broker with different job profiles and Grid middleware and conclude with the lessons learnt from our development experience. Copyright © 2007 John Wiley & Sons, Ltd.     

Metadata for Managing Grid Resources in Data Mining Applications

The Grid is an infrastructure for resource sharing and coordinated use of those resources in dynamic heterogeneous distributed environments. The effective use of a Grid requires the definition of metadata for managing the heterogeneity of involved resources that include computers, data, network facilities, and software tools provided by different organizations. Metadata management becomes a key issue when complex applications, such as data-intensive simulations and data mining applications, are executed on a Grid. This paper discusses metadata models for heterogeneous resource management in Grid-based data mining applications. In particular, it discusses how resources are represented and managed in the Knowledge Grid, a framework for Grid-enabled distributed data mining. The paper illustrates how XML-based metadata is used to describe data mining tools, data sources, mining models, and execution plans, and how metadata is used for the design and execution of distributed knowledge discovery applications on Grids.     

EDGeS: Bridging EGEE to BOINC and XtremWeb

Desktop Grids, such as XtremWeb and BOINC, and Service Grids, such as EGEE, are two different approaches for science communities to gather computing power from a large number of computing resources. Nevertheless, little work has been done to combine these two Grid technologies in order to establish a seamless and vast Grid resource pool. In this paper we present the EGEE Service Grid, the BOINC and XtremWeb Desktop Grids. Then, we present the EDGeS solution to bridge the EGEE Service Grid with the BOINC and XtremWeb Desktop Grids.     

Provide Virtual Distributed Environments for Grid computing on demand

Grid users always expect to meet some challenges to employ Grid resources, such as customized computing environment and QoS support. In this paper, we propose a new methodology for Grid computing – to use virtual machines as computing resources and provide Virtual Distributed Environments (VDE) for Grid users. It is declared that employing virtual environment for Grid computing can bring various advantages, for instance, computing environment customization, QoS guarantee and easy management. A light weight Grid middleware, Grid Virtualization Engine, is developed accordingly to provide functions of building virtual environment for Grids. We also present a typical use case, on-demand build a virtual e-Science infrastructure to justify the methodology.     

Online algorithms for advance resource reservations

We consider the problem of providing QoS guarantees to Grid users through advance reservation of resources. Advance reservation mechanisms provide the ability to allocate resources to users based on agreed-upon QoS requirements and increase the predictability of a Grid system, yet incorporating such mechanisms into current Grid environments has proven to be a challenging task due to the resulting resource fragmentation. We use concepts from computational geometry to present a framework for tackling the resource fragmentation, and for formulating a suite of scheduling strategies. We also develop efficient implementations of the scheduling algorithms that scale to large Grids. We conduct a comprehensive performance evaluation study using simulation, and we present numerical results to demonstrate that our strategies perform well across several metrics that reflect both user- and system-specific goals. Our main contribution is a timely, practical, and efficient solution to the problem of scheduling resources in emerging on-demand computing environments.     

Market-oriented Grids and Utility Computing: The State-of-the-art and Future Directions

Traditional resource management techniques (resource allocation, admission control and scheduling) have been found to be inadequate for many shared Grid and distributed systems, that consist of autonomous and dynamic distributed resources contributed by multiple organisations. They provide no incentive for users to request resources judiciously and appropriately, and do not accurately capture the true value, importance and deadline (the utility) of a user’s job. Furthermore, they provide no compensation for resource providers to contribute their computing resources to shared Grids, as traditional approaches have a user-centric focus on maximising throughput and minimising waiting time rather than maximising a providers own benefit. Consequently, researchers and practitioners have been examining the appropriateness of ‘market-inspired’ resource management techniques to address these limitations. Such techniques aim to smooth out access patterns and reduce the chance of transient overload, by providing a framework for users to be truthful about their resource requirements and job deadlines, and offering incentives for service providers to prioritise urgent, high utility jobs over low utility jobs. We examine the recent innovations in these systems (from 2000–2007), looking at the state-of-the-art in price setting and negotiation, Grid economy management and utility-driven scheduling and resource allocation, and identify the advantages and limitations of these systems. We then look to the future of these systems, examining the emerging ‘Catallaxy’ market paradigm. Finally we consider the future directions that need to be pursued to address the limitations of the current generation of market oriented Grids and Utility Computing systems.     

Fault tolerant and prioritized scheduling in OGSA‐based mobile Grids

Grids and mobile Grids can form the basis and the enabling technology for pervasive and utility computing due to their ability to being open, highly heterogeneous and scalable. In this paper we present a scheme for advancing quality of service (QoS) attributes, such as fault tolerance and prioritized scheduling, in OGSA‐based mobile Grids. The fault tolerance is achieved by producing and managing sufficient replicas of tasks submitted for execution on the mobile Grid resources. We design a simple and efficient prioritization scheme, which allows the scheduling of the tasks submitted by the Grid users as distinguished priorities that can be managed and exploited as a QoS parameter by the Grid infrastructure operator. The results that are presented show the efficiency of the proposed scheme in being simple and additionally enriching with reliability and QoS features the applications that are built on the concept of mobile Grids. Copyright © 2008 John Wiley & Sons, Ltd.