CASANOVA, H., AND DONGARRA, J. NetSolve: A network-enabled server for solving computational science problems. The International Journal of Supercomputer Applications and High Performance Computing 11, 3 (1997), 212--223.  Home/Search   Document Details and Download   Summary   Related Articles   Check

....show conclusively that our data parallel approach scales favorably with data size both in sequential mode where the models are trained sequentially and in parallel where they maximally leverage the presence of multiple processors. The parallel scalability studies have been conducted using NetSolve [21], a computational platform which facilitates grid based heterogenous computing in a transparent and ecient manner. The Southampton Computational Grid was utilized in part to conduct these studies. It is generally recognized that data mining in the future would shift to such grid based computing ....

Casanova, H., Dongarra, J., NetSolve: A Network-enabled Server for Solving Computational Science Problems The International Journal of Supercomputer Applications and High Performance Computing,, Vol. 11(3) pp. 212-223, 1997

....is maintained in the shared objects of each CMM. The resources and services themselves utilize a push protocol for resource dissemination. The AMs schedulers perform resource discovery and scheduling by querying shared objects. 10.12. NetSolve: A Network Enabled Computational Kernel Netsolve [48] is a client agent server paradigm based network enabled application server. It is designed to solve computational science problems in a distributed environment. The Netsolve system integrates network resources including hardware and computational software packages into a desktop application. ....

H. Casanova and J. Dongarra. Netsolve: A network-enabled server for solving computational science problems. Int'l Journal of Supercomputer Applications and High Performance Computing, 11(3):212- 223, Fall 1997.

.... presented in the seminal Grid book [8] Other research e orts have focused on predicting resource availability in the Grid, e.g. in the NWS system [17] The next wave of interest was in particular classes of applications and supporting tools for them, with such important projects as Netsolve [7] and GridPP [13] Algorithmic and programming methodology aspects seem to have been largely neglected at this early stage of Grid research and are therefore not yet properly understood. Initial experience has shown that entirely new approaches to software development and programming are required ....

H. Casanova and J. Dongarra. NetSolve: A network-enabled server for solving computational science problems. Int. J. of Supercomputing Applications and High Performance Computing, 3(11):212223, 1997.

....Then the system cannot forecast the future evolutions of the system, contrary to NWS. Moreover this system is tightly related to the whole Globus system and cannot be used in other projects easily. Finally it can only monitor the network, and not the CPU and memory of the machines. NetSolve [2], used its own set of sensor to monitor the system until the version 1.3. In the current version, NWS is used for the monitoring issues. The routine model is merely expressed by an expression of the form x s y , where s is the size of the parameters, and x, y express the complexity of the ....

H. Casanova and J. Dongarra. NetSolve: A networkenabled server for solving computational science problems. The International Journal of Supercomputer Applications and High Performance Computing, 11(3):212--223, Fall 1997.

....the home site of a proprietary provider. Nevertheless, the authorized user can access the tool via standard Web browsers. Our work is related to other projects that have provided end users with the means to transparently access and use distributed resources. Examples are CCS [1] MMM [2] NetSolve [3], Ninf [4] RCS [5] Rivendell [6] and Schooner [7] We address the second issue by providing, as part of the ParHub, a family of tools that are integrated in that they support a common underlying programming methodology and they can exchange intermediate data. In this regard, our work is ....

....as the Exploratorium [25] and JSPICE [26] are based on scripts that need to be modified in order to add any new application to the system. Other designs are more flexible. The MOL [27] prototype, for example, employs static Web interfaces that can be adapted for individual tools. The NetSolve [3], Ninf [4] and RCS [5] systems are based on structured designs that target numerical software libraries. However, static interfaces are not adequate for all tools, and structured approaches cannot be easily applied to general purpose applications. Another problem is that the majority of these ....

H. Casanova and J. Dongarra. NetSolve: a network enabled server for solving computational science problems. International Journal of Supercomputer Applications, 11(3):212--223, Fall 1997.

....of theses resources. 2) From the system point of view, they should manage the shared resources in a way that preclude unmanageable workloads. The well known Client Broker Server (CBS) software architecture may play a significant role towards theses aims. Distributed environments like NetSovle [8], Ninf [9] or RCS [10] already allow to ease the access and the management of high performance computers. These Metacomputing tools are mostly dedicated to MAN and WAN networks. They address the problems of heterogeneity, reliability and security through the use of standard components (TCP IP ....

H. Casanova and J. Dongarra. NetSolve: A network-enabled server for solving computational science problems. The International Journal of Supercomputer Applications and High Performance Computing, 11(3):212--223, Fall 1997.

....as CORBA and Java, for accessing remote computers and databases. At a 1995 NSF workshop on PSEs [15] the need to develop and evaluate PSE infrastructure and tools was stressed. Subsequently a number of prototype PSEs have been developed. Many early PSEs focus on linear algebra 2 computations [4] and the solution of partial differential equations, and as yet only a few prototype PSEs have been developed especially for science and engineering applications [6, 16] However, this is likely to change over the next few years. Tools for building specific types of PSEs have been developed, such ....

H. Casanova and J. J. Dongarra. NetSolve: A Network-Enabled Server for Solving Computational Science Problems. Int. Journal of Supercomputing Applications, 11(3), 1997. 20

....at implementing prototype PSEs and at developing the software infrastructure or middleware for constructing PSEs. Much of this work has been presented in a recent book by Houstis et al. 19] Initially, many of the prototype PSEs that were developed focused on linear algebra computations [10] and the solution of partial di erential equations [21] More recently prototype PSEs have been developed speci cally for science and engineering applications [12, 13, 22, 33] Tools for building speci c types of PSEs, such as PDELab [35] a system for building PSEs for solving PDEs) and PSEWare ....

H. Casanova and J. J. Dongarra, \NetSolve: A Network-Enabled Server for Solving Computational Science Problems," Int. J. Supercomputing Appl. Vol. 11, No. 3, pp. 212-223, Fall 1997.

....PSEs. Since the 1991 workshop, PSE research has been mainly directed at implementing prototype PSEs and at developing the software infrastructure or middleware for constructing PSEs. Initially, many of the prototype PSEs that were developed focused on linear algebra computations [9] and the solution of partial differential equations [19] More recently prototype PSEs have been developed specifically for science and engineering applications [11, 12, 17, 25] Tools for building specific types of PSEs, such as PDELab [27] a system for building PSEs for solving PDEs) and ....

H. Casanova and J. J. Dongarra, "NetSolve: A Network-Enabled Server for Solving Computational Science Problems," Int. J. Supercomputing Appl. Vol. 11, No. 3, pp. 212-223, Fall 1997.

....are trustworthy. The term service based metacomputer was coined several years ago and is similar to the notion of computing portals. In both cases, computing resources are not made directly available to the Internet world, but indirectly through a set of pre installed applications. Also, Netsolve [5] and Ninf [15] have introduced comparable concepts. The MOLKernel builds upon these ideas, and is meant to provide a more generic infrastructure for setting up a computing portal environment. 4. A Robust Management Architecture The task of the MOL Kernel is to provide the core functionality ....

H. Casanova and J. Dongarra. NetSolve: A network-enabled server for solving computational science problems. The International Journal of Supercomputer Applications and High Performance Computing, 11(3):212--223, 1997.

....the home site of a proprietary provider. Nevertheless, the authorized user can access the tool via standard Web browsers. Our work is related to other projects that have provided end users with the means to transparently access and use distributed resources. Examples are CCS [1] MMM [2] NetSolve [3], Ninf [4] RCS [5] Rivendell [6] and Schooner [7] We address the second issue by providing, as part of the ParHub, a family of tools that are integrated in that they support a common underlying programming methodology and they can exchange intermediate data. In this regard, our work is ....

....as the Exploratorium [25] and JSPICE [26] are based on scripts that need to be modified in order to add any new application to the system. Other designs are more flexible. The MOL [27] prototype, for example, employs static Web interfaces that can be adapted for individual tools. The NetSolve [3], Ninf [4] and RCS [5] systems are based on structured designs that target numerical software libraries. However, static interfaces are not adequate for all tools, and structured approaches cannot be easily applied to general purpose applications. Another problem is that the majority of these ....

H. Casanova and J. Dongarra. NetSolve: a network enabled server for solving computational science problems. International Journal of Supercomputer Applications, 11(3):212--223, Fall 1997.

....7. 2 Related Work There is a large amount of related work which has made collaborative use of computational resources over a global network, including low level communication systems such as MPI [8] and PVM [24] and higher level dedicated systems, including Globus [9] Legion [12] and NetSolve [4]. Although these systems offer heterogeneous collaboration of multiple systems in parallel some of them in wide area setting they involve rather complex maintenance of different binary code, multiple execution environments, etc. CORBA [19] defines a middleware that bridges distributed ....

H. Casanova and J. Dongarra. NetSolve: A network-enabled server for solving computational science problems. The International Journal of Supercomputer Applications and High Performance Computing, 11(3):212--223, Fall 1997.

....models and frameworks that will support the development of distributed applications on the Web. Such an example is WebFlow [40] a visual programming environment for Web Java based coarse grain distributed computing. Two additional independent (but very similar) projects are Ninf [41] and NetSolve [42], which aim to enable users to access computational resources including hardware, software and scientific data distributed across a wide area network. Computational resources are shared as remote libraries, while a central server maintains global resource information regarding computational ....

....respect, they do not qualify for straightforward deployment of information processing services in today s business Intranets and Extranets. To meet such needs, application tailored environments have been developed [38, 39] as well as more generic frameworks for distributed processing applications [40,41,42]. However, such approaches in general address the problem area in a rather centralized way and do not account for management of complex processing tasks or for resource allocation and pricing. Thus, such projects should be regarded as approaches to distributed information processing, but not as ....

H. Casanova and J. Dongarra, "NetSolve: A Network-Enabled Server for Solving Computational Science Problems", The International Journal of Supercomputer Applications and High Performance Computing, vol. 11(3), 212-223, 1997.

....done by a local frontend which has a mathematical model of the temporal behaviour for every backend. Therefore, the integration of a new backend results in a change of the frontend. The Netsolve system supports special interfaces, e.g. to MatLab or C, instead of one general interface like CORBA [CD97] Load balancing is done via special agents, which are more complex than traders. The system Ninf is close to our approach [SSN 96] It also uses distributed and heterogeneous computers which are administrated by metaservers comparable with traders. But instead of our object oriented, open ....

H. Casanova and J. Dongarra. NetSolve: A Network-Enabled Server for Solving Computational Science Problems. The International Journal of Supercomputer Applications and High Performance Computing, 11(3):212--223, 1997. 4

....in the type of services. ReGTime [4] only supports the execution of PVM applications, UNICORE and the Hypercomputer [20] are restricted to general batch systems. DISCWorld [13] offers general services like Amica but is not based on the factory pattern. Some projects, e.g. RCS [2] or NetSolve [3], provide a library which is linked to an application and encapsulates a run time system which uses distributed resources to compute a library function. A disadvantage of this approach is the difficulty of updating the library. Data Storage and Communication In Amica, communication is done by ....

H. Casanova and J. Dongarra. NetSolve: A network-enabled server for solving computational science problems. The International Journal of Supercomputer Applications and High Performance Computing, 11(3):212--223, 1997.

....techniques allow a same application to access and use various remote resources ( 7] In order to refer our project through the Metacomputing community, we brie y present related projects or general purpose communication schemes which have been already developed. The NetSolve project [2] aims at bringing remote access to scienti c computing libraries. But, even if our goal is also to o er access to remote libraries, the nature of the data (images) and the processing tools are very di erent. Our approach is speci cally oriented to medical image processing and is not suited for ....

H. Casanova and J. Dongarra. NetSolve: A Network-Enabled Server for Solving Computational Science Problems. The International Journal of Supercomputer Applications and High Performance Computing, 11(3):212-223, 1997.

.... at the cost of rewriting applications to fit the project libraries and languages [9] The Globus project also aims at providing a global metacomputing object oriented environment, but is oriented towards a different set of problems than Aglets H SWEB and is non Java based [8] Projects in [5, 7, 6] are working on global computing software infrastructures, but are generally still in the design stage. Scheduling issues in heterogeneous computing for large scale scientific applications using network bandwidth and load information are addressed in the non Java based AppLeS project [4] Recent ....

H. Casanova and J. Dongarra. NetSolve: A network-enabled server for solving computational science problems. The International Journal of Supercomputer Applications and High Performance Computing, 11(3):212--223, Fall 1997.

....library should provide startFFT( and finishFFT( calls, which the programmer can use to structure the calling application as needed. While SCaLAPACK does not provide asynchronous operations, the NetSolve interface provides asynchronous access to numerical library routines for distributed systems [34]. Our results reinforce the benefits of asynchronous entry points to numerical libraries. The current KeLP model does not explicitly support block cyclic data layouts. The LU application simulated a block cyclic layout as a multi block layout, assigning many blocks per node. We conclude that ....

H.Casanova and J.Dongarra, "NetSolve: a Network Enabled Server for Solving Computational Science Problems," Int. Journal of Supercomputer Applications and High Performance Computing, Vol. 11, Fall 1997, pp. 212--223.

....is a WWW interface to the software and computational resources at the OTC. NEOS currently solves usual and stochastic problems and linear network optimisation problems. The home page also points to a decision tree for optimisation problems and software. 2. 9 NetSolve NetSolve [10, 34] uses a client server agent software architecture with Condor for its distributed computing management. NetSolve is intended to provide transparent access to a a whole variety of software libraries, highly tuned for the target architecture. This improves maintainability of software and avoids the ....

H Casanove and J Dongarra. Netsolve: A network-enabled server for solving computational science problems. Int. J. Supercomputer Applications and High-performance Computing, 11:212--23, 1997.

....collaborative computing: Scientists collaborating on research projects can make use of the proposed architecture to facilitate the exchange and processing of massive data files. By combining the network storage model embodied in IBP with the availability of computational servers such as NetSolve [CD97], scientists can have access to computational services with little or no administrative overhead. As an example, a scientist can use IBP MIME to send a huge uncompressed video file to a colleague, who can instruct the local IBP MIME handler to forward the incoming file directly to to an ....

Henry Casanova and Jack Dongara. NetSolve: A Network-Enabled Server for Solving Computational Science Problems. International Journal of Supercomputer Applications and High Performance Computing, 11(3), Fall 1997.

....library should provide startFFT( and finishFFT( calls, which the programmer can use to structure the calling application as needed. While SCaLAPACK does not provide asynchronous operations, the NetSolve interface provides asynchronous access to numerical library routines for distributed systems [34]. Our results reinforce the benefits of asynchronous entry points to numerical libraries. The current KeLP model does not explicitly support block cyclic data layouts. The LU application simulated a block cyclic layout as a multi block layout, assigning many blocks per node. We conclude that KeLP ....

H. Casanova and J. Dongarra, "NetSolve: a network enabled server for solving computational science problems," Int. Journal of Supercomputer Applications and High Performance Computing, vol. 11, pp. 212--23, Fall 1997.

....library should provide startFFT( and finishFFT( calls, which the programmer can use to structure the calling application as needed. While ScaLAPACK does not provide asynchronous operations, the NetSolve interface provides asynchronous access to numerical library routines for distributed systems [41]. The results from this dissertation further reinforce the benefits of asynchronous entry points to numerical libraries, 149 even in more tightly coupled cluster environments. We note that the current KeLP model does not explicitly support blockcyclic data layouts. The LU application simulated a ....

N. Casanova and J. Dongarra. NetSolve: a network enabled server for solving computational science problems. International Journal of Supercomputer Applications and High Performance Computing, 11(3):212--23, Fall 1997. 162

.... load balancing environment, at the cost of rewriting applications to fit the project libraries and languages [18] The Globus project also aims at providing a global metacomputing object oriented environment, but is oriented towards a different set of problems than H SWEB [14] Projects in [8, 11, 9] are working on global computing software infrastructures, but are generally still in the design stage. Scheduling issues in heterogeneous computing for large scale scientific applications using network bandwidth and load information are addressed in [7] The above work deals with an integration ....

H. Casanova and J. Dongarra. NetSolve: A network-enabled server for solving computational science problems. The International Journal of Supercomputer Applications and High Performance Computing, 11(3):212--223, Fall 1997.

....computing environments. But it is unrealistic to expect the transition of research professionals to the Grid to be anything but halting and slow if it means abandoning the scientific computing environments that they rightfully view as a major source of their productivity. The NetSolve project [1] addresses this difficult problem directly: the purpose of NetSolve is to create the middleware necessary to provide a seamless bridge between the simple, standard programming interfaces and desktop systems that dominate the work of computational scientists, and the rich supply of services ....

Casanova, H. and Dongarra, J. (1997) NetSolve: A network-enabled server for solving computational science problems. The International Journal of Supercomputer Applications and High Performance Computing, 11(3), 212 -- 223.

No context found.

CASANOVA, H., AND DONGARRA, J. NetSolve: A network-enabled server for solving computational science problems. The International Journal of Supercomputer Applications and High Performance Computing 11, 3 (1997), 212--223.

No context found.

H. Casanova and J. Dongarra. NetSolve: a network enabled server for solving computationa l science problems. The International Journal of Supercomputer Applications and H igh Performance Computing, 11(3):212--223, 1997.

No context found.

H. Casanova and J. Dongarra. NetSolve: A Network-Enabled Server for Solving Computational Science Problems. International Journal of Supercomputer Applications and High Performance Computing, 11(3):212 -- 213, Fall 1997.

No context found.

CASA97 Casanova, H., Dongarra, J., NetSolve: A Network-Enabled Server for Solving Computational Science Problems, The International Journal of Supercomputer Applications and High Performance Computing, Vol. 11, No. 3, pp. 212-223, Fall 1997.

No context found.

H. Casanova and J. Dongarra, NetSolve: A network-enabled server for solving computational science problems. The Int. J. Supercomputer Applications and High Performance Computing, 11 (3): 212-223, 1997.

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC