Catálogo de publicaciones - libros

Grid computing has gone through some generations and as a result only a few widely used middleware architectures remain. Using the tools of these middlewares different resource brokers have been developed to automate job submission over different grids. As grid resources were grouped to Virtual Organizations, users seem to become isolated by these groups. Enhancing interoperability among these VOs and grids will be the main issue of future generation grids. This paper describes a meta-brokering architecture that shows how to enable the interoperability of various grids through their own resource brokers.

Submission of multiple jobs in a distributed and heterogeneous environment is required by applications that rely on the “public-resource computing” paradigm. We present here a scientific scenario for the analysis of astronomical data, where some nodes are responsible for maintaining and advertising job description files and other so called worker nodes, are dispersed over the Grid to execute the jobs. Job assignment is performed through a mechanism that matches adverts, containing job descriptions, with job queries that are sent by available workers across the Grid exploiting an underlying super-peer topology. With an analogous mechanism, a worker locates the input data file needed to run a job and downloads it from a data center node. This paper presents a super-peer protocol for the submission of a very large number of jobs on a Grid environment. The super-peer architecture enables the replication of data files on multiple data centers, which helps reduce the processing load and speed up the application. A simulation analysis has been performed to evaluate the impact of application and network parameters on performance results.

This paper describes a scheduling algorithm used to execute parallel and distributed applications on a Global Computing (GC) environment, called XtremWeb-CH ( XWCH ). XWCH is an improved version of a GC tool called XtremWeb ( XW ). XWCH is an enrichment of XW allowing it to match P2P concepts: distributed scheduling, distributed communication and development of symmetrical models. The scheduling algorithm takes into account the heterogeneity and volatility of nodes. This paper illustrates the performance of XWCH in a real CPU time consuming application.

Grid computing has gone through some generations and as a result only a few widely used middleware architectures remain. The Globus Toolkit is the most widespread middleware in most of the current production grid systems, but the LCG-2 middleware dominates in Europe. The paper describes a brokering solution that enables the interoperability of various Globus and LCG-2 based grids during the execution of workflow applications, and supports users to utilize computing and storage resources from multiple production grids by a single application. The development and execution of such applications can be managed by a Web-based Grid portal called P-GRADE Portal, and the brokering of the workflows is carried out by its integrated GTbroker and LCG-2 broker component.

Among existing grid middleware approaches, one simple, powerful, and flexible approach consists of using servers available in different administrative domains through the classic client-server or Remote Procedure Call (RPC) paradigm. Network Enabled Servers (NES) implement this model also called GridRPC. Clients submit computation requests to a scheduler whose goal is to find a server available on the grid. The aim of this paper is to give an overview of an NES middleware developed in the GRAAL team called DIET and to describe recent developments. DIET (Distributed Interactive Engineering Toolbox) is a hierarchical set of components used for the development of applications based on computational servers on the grid.

Application deployment is becoming an increasingly hard task, as complex, component-based Grid applications have to be deployed on heterogeneous and dynamic Grids, interfacing to several different component frameworks and Grid middlewares. We describe the architecture of the Grid Execution Agent (GEA), the deployment and resource brokering tool of the Grid.it project. GEA has been designed to ease the deployment of complex Grid applications written in a high-level, structured way. To easily handle different component models over heterogeneous Grid resources, the GEA design exploits multiple levels of abstraction. Our approach allows consistent translation of the high-level requirements from heterogeneous, multi-component applications, to low-level operations over different middlewares. GEA architecture provides a unified interface with services to locate resources, devise initial mapping, and instantiate applications, and it is extensible to new component models. It supports dynamically reconfiguring, self-adapting applications by allowing execution-time resource allocation changes.

In this paper, we present our work on building a Grid information knowledge base, which is a key component of a semantic Grid information system. A Core Grid Ontology (CGO) is developed for building a Grid knowledge base; and the SPARQL query language is adopted to query the knowledge base.

The UNICORE Grid technology provides a seamless, secure, and intuitive access to distributed Grid resources. UNICORE is a full-grown and well-tested Grid middleware system, which today is used in daily production worldwide. Beyond this production usage, the UNICORE technology serves as a solid basis in many European and International projects. In order to foster these ongoing developments, UNICORE is available as open source under BSD licence at http://www.unicore.eu.

This paper builds on extensive experience with the UNICORE middleware to derive requirements for the next generation of Grid execution management systems. We present some well-known architectural ideas and design principles that allow building Grid servers that are adaptable to any type of target systems, from single workstations or PCs to huge supercomputers, and flexible enough for the novel usage scenarios and business models that are coming up in next-generation Grid systems. These ideas are used to implement an execution management system similar in scope to the UNICORE NJS.

The development of UNICORE started as a Grid-enabling middleware with a monolithic security policy that restricted Grid activities to a set of users whose credentials (X.509 certificates) are pre-recorded in a UNICORE User Database (UUDB), and to a task distribution completely defined at job-submission time because the sub-jobs have to be signed by the user with his private key. Later on projects aiming at allowing a restricted interoperability with other Grid middleware lead to the adoption of more flexible approaches like the the Explicit Trust Delegation (ETD). ETD involves implicitly a more general concept: That of an attribute or role which is attached to an identified and authenticated entity and which defines the extent of the authorisations granted to that entity by the target resource. Extending this concept to other authorisation-related aspects of Grid computing is today an area of intensive research, that should also be taken up by the UNICORE developers in order to enable the creation of Virtual Organisations (VOs) that are able to take security as seriously as necessary, and to opt for flexibility as much as possible.