The gLite Workload Management System

Cecchi, Marco; Capannini, Fabio; Dorigo, A.; Ghiselli, A.; Giacomini, Francesco; Maraschini, A.; Marzolla, Moreno; Monforte, S.; Pacini, Fabrizio; Luca, Petronzio; Prelz, F.

doi:10.1007/978-3-642-01671-4_24

Cited by 18 publications

(7 citation statements)

References 1 publication

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Resources at 20 more sites are available for members of other VOs. Coordination between LRMs of different sites is done by the gLite Workload Management System (WMS) [43]. A user has to supplement each submitted job with a job description file that specifies the job properties and requirements using JDL, a high level language.…”

Section: Grid Testbedmentioning

confidence: 99%

Passive Network Awareness as a Means for Improved Grid Scheduling

Elkhatib

Edwards

2015

J Grid Computing

View full text Add to dashboard Cite

Grids enable sharing resources of heterogeneous nature and administration. In such distributed systems, the network is usually taken for granted which is potentially problematic due to the complexity and unpredictability of public networks that typically underlie grids. This article introduces GridMAP, a mechanism for considering the network state for enhancing grid scheduling. Network measurements are collected in a passive manner from a user-centric vantage point. This mechanism has been evaluated on a production e-science grid infrastructure, with results showing the ability of GridMAP to improve grid scheduling with minimal network, computational and deployment overheads.

show abstract

Section: Grid Testbedmentioning

confidence: 99%

Passive Network Awareness as a Means for Improved Grid Scheduling

Elkhatib

Edwards

2015

J Grid Computing

View full text Add to dashboard Cite

show abstract

“…The most well-known systems specically tailored for BoT in the Grid context are APST [6], Nimrod/G [7], Condor [8], MyGrid [9], Cigri [10] and Glite Workload Management System [11]. All these infrastructures work on a best-eort basis, with no performance guarantees.…”

Section: Bot Scheduling Infrastructuresmentioning

confidence: 99%

“…In this section, we evaluate the algorithm proposed in Section 5 when using exactly Eq. (11). We call such algorithm the naive algorithm as it is a straightforward application of the distributed Lagrangian optimization proposed in [2] to our context.…”

Section: Performance Evaluation Of the Naive Algorithmmentioning

confidence: 99%

See 1 more Smart Citation

Fair scheduling of bag-of-tasks applications using distributed Lagrangian optimization

Bertin¹,

Hunold²,

Legrand³

et al. 2014

Journal of Parallel and Distributed Computing

View full text Add to dashboard Cite

International audienceLarge scale distributed systems typically comprise hundreds to millions of entities (applications, users, companies, universities) that have only a partial view of resources (computers, communication links). How to fairly and efficiently share such resources between entities in a distributed way has thus become a critical question. Although not all applications are suitable for execution on large scale distributed computing platform, ideal are the Bag-of-Tasks (BoT) applications. Hence a large fraction of jobs in workloads imposed on Grids is made of sequential applications submitted in the form of BoTs. Up until now, mainly simple mechanisms have been used to ensure a fair sharing of resources among these applications. Although these mechanisms are proved to be efficient for CPU-bound applications, they are known to be ineffective in the presence of network-bound applications. A possible answer resorts to Lagrangian optimization and distributed gradient descent. Under certain conditions, the resource sharing problem can be formulated as a global optimization problem, which can be solved by a distributed self-stabilizing supply and demand algorithm. In the last decade, this technique has been applied to design various network protocols (variants of TCP, multi-path network protocols, wireless network protocols) and even distributed algorithms for smart grids. In this article, we explain how to use this technique for fairly scheduling concurrent BoT applications with arbitrary communication-to-computation ratio on a Grid. Yet, application heterogeneity raises severe convergence and stability issues that did not appear in the previous contexts and need to be addressed by non-trivial modifications. The effectiveness of our proposal is assessed through an extensive set of complex and realistic simulations

show abstract

“…Currently, the systems, such as gLite WMS [24] and GridWay [25], that perform the match-making of requirements to allocate the most appropriate resources for the execution of jobs in the Grid, implement their own policy interface and protocols, which are difficult with regards to the interoperability and portability of Grid applications. For example, in Grids, clients can use policies to express their requirements, constraints, and preferences, and organizations to advertise the capabilities that support their resources.…”

Section: Data Model and Support For Groupsmentioning

confidence: 99%

A replicated information system to enable dynamic collaborations in the Grid

Torres

Moltó

Segrelles

et al. 2012

Concurrency and Computation

View full text Add to dashboard Cite

The main advantage of Grid computing over other distributed computing paradigms is its capability to coordinate the access to data and resources in a virtual multi-institutional environment. To this end, the information system plays a decisive role in selecting the services that meet the applications' needs. This paper presents an information system for the Grid that provides transparent and scalable group communication services to standard Grid applications, with the objective of supporting dynamic collaborations that could help address problems that involve only some participants of a virtual organization. In particular, it enables more flexible delivery mechanisms, which allows applications to select the appropriate services before sending their data to the information system. This significantly enhances the protection of data from unauthorized access, and avoids the transmission of unnecessary messages over the network. The proposed information system is based on the use of XML technologies and replication. It introduces several new advanced features that are not currently supported as a whole by any Grid middleware, such as: several entry points to the information, persistent capabilities, support for advanced queries based on XQuery, and support for the industrial standard WS-Policy. The information system has been stress tested under realistic workloads in a Grid infrastructure with 50 sites. Scalability has been evaluated in up to 1000 messages that can be up to 10 KB in size each, updated with a frequency of 5 min.Initiative is a strong collaborative and cooperative link with the Portuguese National Grid Initiative through the IBERGRID agreement. This infrastructure is connected to the European Grid Infrastructure (EGI) [6]. The IBERGRID VO for Life Science applications (life.vo.ibergrid.eu) is an example of VO that has been created in the context of EGI to support the research activity in the life science community. This VO enables international collaboration in medical imaging, bioinformatics, and drug discovery.However, dynamic collaborations present a challenge to Grid information systems, because they do not provide a simple way to define special access rights agreed by part of the members of a VO. For example, in a system like VOMS [7], this problem occurs when two or more groups of the VO agree to collaborate on the achievement of a particular objective. Although this can be tackled through the use of thematic VO subgroups, this approach has the disadvantage of requiring the approval of an administrator of the VO (except for very few services such as file catalogues), and requires the reconfiguration of the infrastructure to support differentiation among groups. Moreover, any interaction with the membership to groups must be done through the VO admin role, but even so, the usage of groups to support temporal collaborations is a widely used feature (e.g., biomed VO, IBERGRID VOs) because users in the VO are normally authorized to access all the resources, regardless of the group that they belong to...

show abstract

The gLite Workload Management System

Cited by 18 publications

References 1 publication

Passive Network Awareness as a Means for Improved Grid Scheduling

Passive Network Awareness as a Means for Improved Grid Scheduling

Fair scheduling of bag-of-tasks applications using distributed Lagrangian optimization

A replicated information system to enable dynamic collaborations in the Grid

Contact Info

Product

Resources

About