Reservoir Engineering industry is very interested in using ever growing dataset with more and more complex physics and detailed models. The key issue still remains running simulations in an acceptable CPU time. Regard...
详细信息
ISBN:
(纸本)9780889866386
Reservoir Engineering industry is very interested in using ever growing dataset with more and more complex physics and detailed models. The key issue still remains running simulations in an acceptable CPU time. Regarding the trend in hardware technologies, the challenge is to improve the efficiency of reservoir simulation software on a large number of processors. New numerical difficulties and performance problems appear when the number of cells and the number of processors are growing. In this paper, we present different approaches to overcome these kinds of problems. We discuss technical choices such like advanced scalable linear solver algorithm, load balancing issue with different domain decomposition strategies, dynamic space criteria, mesh partitioner strategy and parallel solver performance management, flexible 10 strategy from simple file system to more complex parallel file system or database. We have developed and benchmarked these different solutions on published reference large scale problems and actual case studies with several tens millions of cells. We analyse the results and discuss the efficiency of each solution to overcome the scalability difficulties and performance limitations due to load unbalance.
Although originally fostering the principle of free resource sharing, academia has identified the need for charging mechanisms for grid services offered in order to better control resource usage. Within the IT industr...
详细信息
ISBN:
(纸本)9780889866379
Although originally fostering the principle of free resource sharing, academia has identified the need for charging mechanisms for grid services offered in order to better control resource usage. Within the IT industry on the other hand, numerous initiatives are aiming at creating commercial grid services. Both need payment systems but most of the available solutions are bespoke. We present a certificate based payment solution that requires no modification of existing middleware APIs. This transparency guarantees seamless operation with deployed, non-payment aware grid infrastructures and enables hence the simultaneous and interchangeable use of free and paid for grid services. GridBank, the central component in our prototype implementation, both issues and redeems digitally signed cheques. The only common nominator being certificate based authentication allows GridBank to be integrated not only to grid middleware but also to any software infrastructure leveraging X.509 certificates for authentication. We compare our solution to other existing payment solutions for grids, describe its detailed architecture and conclude by analysing its features, strengths and weaknesses.
In a Grid environment the main components that manage the job workflow life are the Grid Resource Layer, the Grid Information System and the Grid Information Data Model. Until now, only some of the known computational...
详细信息
ISBN:
(纸本)088986568X
In a Grid environment the main components that manage the job workflow life are the Grid Resource Layer, the Grid Information System and the Grid Information Data Model. Until now, only some of the known computational resources, such as a hardware machines and a batch queueing system, have been take into account as valid Resource Framework Layer instances. However many types of virtual computing machines exist, such as a Java Virtual Machine, a parallel Virtual Machine and a Data Source Engine. The actual representation of a Grid resource in the Grid Information System and Data Model concerns only hardware computing machines. We argue that a Data Source Engine is a valid instance for a Grid computing model. Then we define a new Grid element that enables the access to a Data Source Engine and a Data Source, totally integrated with the Grid Monitoring and Discovery System and Resource Broker.
Domus is an architecture for distributed Hash Tables (DHTs) tailored to a shared-all cluster environment. Domus DHTs build on a (dynamic) set of cluster nodes;each node may perform routing and/or storage tasks, for on...
详细信息
ISBN:
(纸本)9780889866379
Domus is an architecture for distributed Hash Tables (DHTs) tailored to a shared-all cluster environment. Domus DHTs build on a (dynamic) set of cluster nodes;each node may perform routing and/or storage tasks, for one or more DHTs, as a function of the node base (static) resources and of its (dynamic) state. Domus DHTs also benefit from a rich set of user-level attributes and operations. pDomus is a prototype of Domus that creates an environment where to evaluate the architecture concepts and features. In this paper, we present a set of experiments conduced to obtain figures of merit on the scalability of a specific DHT operation, with several lookup methods and storage technologies. The evaluation also involves a comparison with a database and a P2P-oriented DHT platform. The results are promising, and a motivation for further work.
Cloud computing is attracting considerable attention as a paradigm shift in the way of supplying and using computing resources. To realize a platform of Cloud computing, large-scale data centers equipped with a huge n...
详细信息
ISBN:
(纸本)9780889868113
Cloud computing is attracting considerable attention as a paradigm shift in the way of supplying and using computing resources. To realize a platform of Cloud computing, large-scale data centers equipped with a huge number of physical resources have to be constructed and operated. One of the big concerns is then how to reduce the resource management and operational cost. Moreover, for reasons of security, users require isolation of their virtual systems running on a shared Cloud platform. Especially, business application systems need strong isolation to protect their private and confidential information. Accordingly, we have been developing new technologies to support the operation and management of such next-generation data centers. In this paper, we describe the design and implementation of ITcell system a platform for IaaS (Infrastructure as a Service) style Cloud computing. Our concept for ITcell is treating physical resources like cells in living organisms-numerous, easily replaceable and integrated in a larger whole. In the development of ITcell system, we addressed and solved the following issues: (1) efficient management of physical resources;(2) rapid integration of new physical servers;(3) managing a VAS (virtual application system) as a single unit;and (4) ensuring network isolation between VAS instances on a shared platform.
Adaptive task partitioning may introduce implicit hot spots in the computations which affect the parallel performance of the adaptive multivariate integration algorithm. To maintain parallel efficiency, one needs to b...
详细信息
ISBN:
(纸本)088986392X
Adaptive task partitioning may introduce implicit hot spots in the computations which affect the parallel performance of the adaptive multivariate integration algorithm. To maintain parallel efficiency, one needs to balance the load of the processors by distributing the computationally intensive parts of the problem domain. We study and improve the performance of the scheduler based or centralized load balancing method in the adaptive task partitioning algorithm implemented in ParInt.
In this paper, we propose an algorithm to solve the group k-mutual exclusion (Gk-ME) problem for distributedsystems. The Gk-ME problem is concerned with controlling the concurrent accesses of some resource by at most...
详细信息
ISBN:
(纸本)088986392X
In this paper, we propose an algorithm to solve the group k-mutual exclusion (Gk-ME) problem for distributedsystems. The Gk-ME problem is concerned with controlling the concurrent accesses of some resource by at most k nodes with the constraint that no two distinct resources can be accessed simultaneously. The proposed algorithm utilizes a token circulation mechanism and does not require the nodes to have identifications. The delay and session switches of the proposed algorithm are Ω(n2) and Ω(n), respectively.
The purpose of this case study is to develop a performance model for an enterprise grid for performance management grid capacity planning(1). The target environment includes,grid applications such as health-care arid ...
详细信息
ISBN:
(纸本)9780889866386
The purpose of this case study is to develop a performance model for an enterprise grid for performance management grid capacity planning(1). The target environment includes,grid applications such as health-care arid financial services where the data is located primarily within the resources of a worldwide corporation. The approach is to build a discrete event simulation model for a representative work-flow grid. Five work-flow classes, found using a customized k-means clustering algorithm characterize the workload of the grid. Analyzing the gap between the simulation arid measurement data validates the model. The case study demonstrates that the simulation model can be used to predict the V a given a workload forecast. The,rid system performance model is also used to evaluate alternative scheduling strategies. The simulation model is flexible arid easily incorporates several system details.
Sombrero is an object-oriented single address space operating system whose virtual address space can be distributed over multiple nodes. This paper presents recent additions to the design and implementation of the low...
详细信息
ISBN:
(纸本)9780889866386
Sombrero is an object-oriented single address space operating system whose virtual address space can be distributed over multiple nodes. This paper presents recent additions to the design and implementation of the lower level architecture of Sombrero. This design exploits Sombrero's object-oriented development model to provide an implementation of a lower-level architecture for the Alpha platform. Software representations of the hardware system's major components have been implemented. The existing middle level architectural modules now use the lower-level services, These include board and processor level hardware interfaces as well as interrupt processing, device driver and protocol component architectures. We contrast differences in the ways these things are designed and operate in a single address space environment with the way they are done in a conventional multiple address space operating system. Features covered include the use of passive servers, simple efficient thread switching and threads blocking within interrupts. Passive and active device drivers are described and the rational for each mode given. The performance of the Sombrero protocol stack was closely examined with attention to its passive service architecture and thread switching method. The current Sombrero Prototype is described.
Mutual exclusion is a fundamental problem in distributed processing systems. A generalization of mutual exclusion called k-exclusion for shared memory systems was introduced by Fischer et al. in [8] and subsequently s...
详细信息
Mutual exclusion is a fundamental problem in distributed processing systems. A generalization of mutual exclusion called k-exclusion for shared memory systems was introduced by Fischer et al. in [8] and subsequently studied in [1, 2, 3, 5]. In this paper, we present a simple solution to this problem and prove its correctness. Our solution is efficient both in time and space.
暂无评论