Grid'5000: a large scale and highly reconfigurable grid experimental testbed

Cappello, Franck; Caron, Eddy; Daydé, Michel; Desprez, Frédéric; Jégou, Yvon; Primet, Pascale Vicat-Blanc; Jeannot, Emmanuel; Lanteri, Stéphane; Leduc, Julien; Melab, Nouredine; Mornet, Guillaume; Namyst, Raymond; Quétier, Benjamin; Richard, Olivier

doi:10.1109/grid.2005.1542730

Cited by 186 publications

(104 citation statements)

References 15 publications

Supporting

Mentioning

102

Contrasting

Unclassified

Order By: Relevance

“…In all of the following benchmarks, we measured the total network traffic by using an instrumented local SOCKS server [9] on every machine. All JVMs are instructed to forward all of their connections to the local SOCKS server which then simply forwards the connection and prints its transferred size at the end.…”

Section: Methodsmentioning

confidence: 99%

Garbage Collecting the Grid: A Complete DGC for Activities

Caromel

Chazarain

Henrio

2007

Lecture Notes in Computer Science

View full text Add to dashboard Cite

Abstract. Grids are becoming more and more dynamic, running parallel applications on large scale and heterogeneous resources. Explicitly stopping a whole distributed application is becoming increasingly difficult. In that context, there is a strong need to free resources a soon as they become useless, leading to automatic termination, using distributed garbage collecting techniques. We propose in this paper a new distributed garbage collector for active objects taking into account cycles but with a complexity similar to the distributed garbage collector of Java/RMI. The algorithm is based on a different approach to collect acyclic and cyclic garbage. On one hand, acyclic garbage is collected by knowing the immediate referencers of an active object and detecting the lack of these referencers. This behavior with respect to acyclic garbage is common to the distributed garbage collector of RMI. On the other hand, cyclic garbage is detected by considering the recursive closure of all the referencers of an active object and finding cycles of active objects waiting for requests. These cycles are found by letting idle active objects make a consensus on a common final activity. The algorithm is fully distributed and has been implemented with no modifications to the local garbage collector. Benchmarks have shown the scalability of the algorithm in a grid context.

show abstract

Section: Methodsmentioning

confidence: 99%

Garbage Collecting the Grid: A Complete DGC for Activities

Caromel

Chazarain

Henrio

2007

Lecture Notes in Computer Science

View full text Add to dashboard Cite

show abstract

“…In order to test our XCP-TCP fairness, we implemented our propositions in the ns2 XCP modules provided by D. Katabi (http://www.ana.lcs.mit.edu/ dina/XCP/) In our simulations, we will focus on 2 different scenario: national small distance Grid (such as the French Grid5000 [4] infrastructure) with an average 20 ms RTT and larger scale Grids with 100 ms RTT. The topology used to test our XCP-TCP fairness solutions will be the one shown in Figure 1.…”

Section: Validating Our Xcp-tcp Fairness Solutionmentioning

confidence: 99%

Lightweight Fairness Solutions for XCP and TCP Cohabitation

Pacheco

Lefèvre

Pham

2008

NETWORKING 2008 Ad Hoc and Sensor Networks, Wireless Networks, Next Generation Internet

View full text Add to dashboard Cite

Abstract. XCP is a promising router-assisted protocol due to its high performance and intra-protocol fairness in fully XCP networks. However, XCP is not inter-operable with current E2E protocols resulting in a poor performance of XCP flows when the resources are shared with TCP-like flows. In this paper, we propose a lightweight fairness solution, that can benefit the cohabitation between XCP and TCP in large bandwidth x delay product networks and long-life flows scenarios, as it is shown in our simulation results.

show abstract

“…Therefore we first needed to analyze a series of configurations involving the page size of a BLOB in which data is stored and how varying sizes of data affect the overhead of read and write operations. We performed tests by deploying BlobSeer on the Grid'5000 scientific instrument [5] across 20 virtual machines located at three different sites. Figure 2 presents the performance times for aggregating objects of varying dimensions (from 1KB to 64MB) with different setups of the DDAS, that is a variation of the Page Size set in BlobSeer when writing and reading data.…”

Section: Test Scenarios and Evaluationmentioning

confidence: 99%

A formal method for rule analysis and validation in distributed data aggregation service

Serbanescu¹,

Pop²,

Cristea³

et al. 2015

World Wide Web

View full text Add to dashboard Cite

The usage of Cloud Serviced has increased rapidly in the last years. Data management systems, behind any Cloud Service, are a major concern when it comes to scalability, flexibility and reliability due to being implemented in a distributed way. A Distributed Data Aggregation Service relying on a storage system meets these demands and serves as a repository back-end for complex analysis and automatic mining of any type of data. In this paper we continue our previous work on data management in Cloud storage. We present a formal approach to express retrieval and aggregation rules with a compact, yet powerful tool called Rule Markup Language. Our extended solution proposes a standard form to schemes and uses the tool to match the rules to the XML form of the structured data in order to obtain the unstructured entries from BlobSeer data storage system. This allows the Distributed Data Aggregation Service (DDAS) to bypass several steps when processing a retrieval request. Our new architecture is more loosely-coupled with a separate module, the new tool, used for transforming the XML entries to standard XML files which represent the final result. We model the dynamic behavior of the system using this new standard to ensure a simpler and efficient representation of the operations performed by the client while maintaining the constraints imposed by a distributed system running in the Cloud. Furthermore we prove that this method correctly performs the translation between the storage model's unstructured view of data and the client's structured objects.

show abstract

Grid'5000: a large scale and highly reconfigurable grid experimental testbed

Abstract: Abstract

Cited by 186 publications

References 15 publications

Garbage Collecting the Grid: A Complete DGC for Activities

Garbage Collecting the Grid: A Complete DGC for Activities

Lightweight Fairness Solutions for XCP and TCP Cohabitation

A formal method for rule analysis and validation in distributed data aggregation service

Contact Info

Product

Resources

About