Yves Caniou

Simbatch: An API for Simulating and Predicting the Performance of Parallel Resources Managed by Batch Systems

In this paper, we describe Simbatch, an API which offers core functionalities to realistically simulate parallel resources and batch reservation systems. The objective is twofold: proposing at the same time a tool to efficiently predict parallel resources usage based on their simulations, and to realistically study Grid scheduling heuristics that may be embedded in a Grid middleware or in a tool that deploys it. Indeed, such predictions can be used in a Grid middleware both for scheduling purposes, and to dynamically tune moldable applications in function of the load of the chosen parallel resource in place of the Grid user. Simbatch simulation experiments show an average error rate under 2% compared to real life experiments conducted with the OAR batch manager.

Diet: new developments and recent results

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.

Large-scale parallelism for constraint-based local search: the costas array case study

We present the parallel implementation of a constraint-based Local Search algorithm and investigate its performance on several hardware platforms with several hundreds or thousands of cores. We chose as the basis for these experiments the Adaptive Search method, an efficient sequential Local Search method for Constraint Satisfaction Problems (CSP). After preliminary experiments on some CSPLib benchmarks, we detail the modeling and solving of a hard combinatorial problem related to radar and sonar applications: the Costas Array Problem. Performance evaluation on some classical CSP benchmarks shows that speedups are very good for a few tens of cores, and good up to a few hundreds of cores. However for a hard combinatorial search problem such as the Costas Array Problem, performance evaluation of the sequential version shows results outperforming previous Local Search implementations, while the parallel version shows nearly linear speedups up to 8,192 cores. The proposed parallel scheme is simple and based on independent multi-walks with no communication between processes during search. We also investigated a cooperative multi-walk scheme where processes share simple information, but this scheme does not seem to improve performance.

Constraint-based local search for the costas array problem

The Costas Array Problem is a highly combinatorial problem linked to radar applications. We present in this paper its detailed modeling and solving by Adaptive Search, a constraint-based local search method. Experiments have been done on both sequential and parallel hardware up to several hundreds of cores. Performance evaluation of the sequential version shows results outperforming previous implementations, while the parallel version shows nearly linear speedups w.r.t. the sequential one, for instance 120 for 128 cores and 230 for 256 cores.

Parallel Local Search for the Costas Array Problem

The Costas Array Problem is a highly combinatorial problem linked to radar applications. We present in this paper its detailed modeling and solving by Adaptive Search, a constraint-based local search method. Experiments have been done on both sequential and parallel hardware up to several hundreds of cores. Performance evaluation of the sequential version shows results outperforming previous implementations, while the parallel version shows nearly linear speedups up to 8,192 cores.

Performance analysis of parallel constraint-based local search

We present a parallel implementation of a constraint-based local search algorithm and investigate its performance results for hard combinatorial optimization problems on two different platforms up to several hundreds of cores. On a variety of classical CSPs benchmarks, speedups are very good for a few tens of cores, and good up to a hundred cores. More challenging problems derived from reallife applications (Costas array) shows even better speedups, nearly optimal up to 256 cores.

Cosmological Simulations using Grid Middleware

Large problems ranging from numerical simulation can now be solved through the Internet using grid middleware. This paper describes the different steps involved to make available a service in the DIET grid middleware. The cosmological RAMSES application is taken as an example to detail the implementation. Furthermore, several results are given in order to show the benefits of using DIET, among which the transparent usage of numerous clusters and a low overhead (finding the right resource and submitting the computing task).

Meta-scheduling and Task Reallocation in a Grid Environment

In this paper, we study the impact of task reallocations/migrations on a grid platform, composed of parallel resources, each supervised by a batch scheduler. Results of simulations using real life traces show that gains of 40% can be attained on the average waiting time of jobs even if tasks are initially mapped by a meta-scheduler on the cluster giving the best completion time.

Toward an international sparse linear algebra expert system by interconnecting the ITBL computational Grid with the Grid-TLSE platform

Complex optimization problems are of high interest for Process Systems Engineering. The selection of the relevant technique for the treatment of a given problem has already been studied for batch plant design issues. Classically, most works reported in the dedicated literature yet considered item sizes as continuous variables. In a view of realism, a similar approach is proposed in this paper, with discrete variables for representing equipment capacities, which leads to a combinatorial problem. For this purpose, a Genetic Algorithm was used, which is multiparametric by nature and a grid approach is perfectly relevant to this case study, since the GA code must be run several times, with different values of some input parameters, to guarantee its stochastic nature. This paper is devoted to the presentation of a grid-oriented GA methodology. Some significant results are highlighted and discussed.In the present paper, the methodology of interoperability between ITBL and Grid-TLSE is described. Grid-TLSE is an expert web site to provides user assistance in choosing the right solver for its problems and appropriate values for the control parameters of the selected solve. The time to solution of linear equation solver strongly depends on the type of problem, the selected algorithm, its implementation and the target computer architecture. Grid-TLSE uses the Diet middleware to distribute computing tasks over the Grid. Therefore, extending the variety of computer architecture by Grid middleware interoperability between Diet and ITBL has a beneficial impact to the expert system. To show the feasibility of the methodology, job transfering program as a special service of Diet was developed.

Transparent Collaboration of GridRPC Middleware using the OGF Standardized GridRPC Data Management API

This paper presents a basic implementation of the OGF standard GridRPC Data Management API and its integration in two different middleware, respectively DIET and Ninf. We have conducted several experiments, showing the benefits a Grid user can expect 1) in terms of computation feasibility and resource usage compared to the current GridRPC context since useless transfers can be avoided; 2) in terms of reducing the completion time of an application to obtain results the soonest (data and temporary results can be easily kept on the computational and storage servers); 3) in terms of code portability, since we show with these examples that at last the same GridRPC code can be compiled and executed within two different GridRPC middleware which implements the GridRPC data management API; 4) finally we thus obtain middleware interoperability without any explicit glue: we show as a proof of concept that resources dispatched across different administrative domains can be used altogether without the underlying distributed data management but this needs knowledge of the topology and/or computing resources: computational servers of DIET and Ninf transparently collaborate to the same calculus by sharing GridRPC data!