J. Pruyne and M. Livny. A Worldwide Flock of Condors : Load Sharing among Workstation Clusters . Journal on Future Generations of Computer Systems, 12, 1996.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....substrate connecting its components (with routers acting as specialpurpose elements invisible in the architecture) while network servers provide all access to storage and computation. Illustrations of such servers and services are plentiful: FTP, NFS, and AFS [7] provide access to storage; Condor [8], NetSolve [9] Ninf [10] provide lightweight access to processing; HTTP provides access to both; GRAM [11] provides access to heavyweight computing resources; LDAP provides access to directory services; and so on. What is notable about these instances, and is equally true of almost all the other ....

J. Pruyne and M. Livny, "A worldwide flock of condors : Load sharing among workstation clusters," Journal on Future Generations of Computer Systems, vol. 12, 1996.

....types of numerical software available on the computational servers become more and more diverse, the resource broker embedded in the agent will need to become increasingly sophisticated, leading to many open research questions. 2 An Interface to the Condor System 2. 1 Overview of Condor Condor [15, 16, 17], developed at the University of Wisconsin, Madison, is a high throughput computing environment that can manage very large collections of distributively owned workstations. Its development has been motivated by the ever increasing need for scientists and engineers to exploit the capacity of such ....

....Condor pool to that job. Once the job has been started, it is periodically checkpointed. It can then be interrupted and migrated within the Condor pool until completion. This organization is partly depicted in Figure 2.2. More details on the Condor system and the software layers can be found in [15, 16, 17]. Interfacing NetSolve and Condor makes it possible to give priority to the local users and provide only underutilized CPU cycles to NetSolve users. 10 2.2 A Condor Pool as a NetSolve Resource Condor Central Manager Negociator Collector Startd Schedd Machine 1 Startd Schedd NetSolve ....

J. Pruyne and M. Livny. A Worldwide Flock of Condors : Load Sharing among Workstation Clusters. Journal on Future Generations of Computer Systems, 12, 1996.

....resources fC j g j=1; k that are accessible to the user via k distinct network links. This is a logical topology, and this work does not attempt to take into account the actual physical network topology of the Grid. Our intent is to model a wide area system, such as a Worldwide Flock of Condors [24] for instance. Each cluster contains a certain number of hosts where a host can be any computing platform, from a single processor workstation to an MPP system, and is available for computation. From now on, we call Storage Storage Storage Host Cluster Network Link User s host and storage ....

J. Pruyne and M. Livny. A Worldwide Flock of Condors : Load Sharing among Workstation Clusters . Journal on Future Generations of Computer Systems, 12, 1996.

....algorithmic and hardware considerations in a metacomputing setting, especially when different numerical resources, or even entire frameworks are integrated into NetSolve. Such integrations are described in the following sections. 2 An Interface to the Condor System 2. 1 Overview of Condor Condor [18, 19, 20], developed at the University of Wisconsin, Madison, is a high throughput computing environment that can manage very large collections of distributively owned workstations. Its development has been motivated by the ever increasing need for scientists and engineers to exploit the capacity of such ....

....the Condor pool to that job. Once the job has been started, it is periodically checkpointed. It can then be interrupted and migrated within the Condor pool until completion. This organization is partly depicted in Figure 3. More details on the Condor system and the software layers can be found in [18, 19, 20]. In the next section, we explain how a Condor pool is being used as the back end to a NetSolve computational resource. 2.2 A Condor Pool as a NetSolve Resource Condor Central Manager Negociator Collector Startd Schedd Machine 1 Startd Schedd NetSolve computational module Machine N Startd Schedd ....

J. Pruyne and M. Livny. A Worldwide Flock of Condors : Load Sharing among Workstation Clusters. Journal on Future Generations of Computer Systems, 12, 1996.

....a novel implementation of this algorithm to increase its scalability. New technologies for NOWs, such as active messages [12] and the SHRIMP project [2] offer the promise of increasing the effective bandwidth available to applications. Advances in operating systems for NOWs such as Condor [4] will improve the performance of applications during times when some workstations on the network are being used intensively. 5 10 15 20 25 0 2 4 6 8 10 12 14 16 Number of processors Mflops per processor LE3 130449 unk. linear elements NOW2 (155Mbps ATM) 130449 unk. linear elements NOW1 ....

D. Epema, M. Livny, R. van Dantzig, X. Evers, and J. Pruyne, A worldwide flock of condors: Load sharing among workstation clusters, Journal on Future Generations of Computers Systems, 12 (1996).

....job with mpirun. Adding a new customized server is rather standard. The procedure consists in creating a new sub directory in . src Server that implements (i) the function that spawns the computation process, ii) the main function of the computation process. For instance in the case of the Condor [17, 18, 19] server, the spawning is done by issuing a call to the condor submit executable and waiting for that call to complete, whereas the main computational function has to read its input from files rather than from the network. Once those two functions have been created, it just suffices to modify the ....

J. Pruyne and M. Livny. A Worldwide Flock of Condors : Load Sharing among Workstation Clusters . Journal on Future Generations of Computer Systems, 12, 1996.

....Rank = 10 true : Rank 0 LoadAvg 0.3 KeyboardIdle 15 60 : DayTime 8 60 60 DayTime 18 60 60; Figure 1. A classad describing a workstation A classad is a mapping from attribute names to expressions. 1 For example, Figure 1 shows a classad that describes a workstation in a Condor [10, 3] pool at the University of Wisconsin, 2 and Figure 2 shows a classad that describes a job submitted for execution. Attributes may be simple integer, real, or string constants, or they may be more complicated expressions constructed with arithmetic and logical operators and record and list ....

D. Epema, M. Livny, R. van Dantzig, X. Evers, and J. Pruyne. A Worldwide Flock of Condors : Load Sharing among Workstation Clusters. Journal on Future Generations of Computer Systems, 12, 1996.

No context found.

J. Pruyne and M. Livny. A Worldwide Flock of Condors : Load Sharing among Workstation Clusters . Journal on Future Generations of Computer Systems, 12, 1996.

No context found.

J. Pruyne and M. Livny. A Worldwide Flock of Condors : Load Sharing among Workstation Clusters . Journal on Future Generations of Computer Systems, 12, 1996.

No context found.

J. Pruyne and M. Livny. A Worldwide Flock of Condors : Load Sharing among Workstation Clusters . Journal on Future Generations of Computer Systems, 12, 1996. 13

No context found.

D.H.JEpema,M.Livny,R.V.Dantzig, X. Evers, and J. Pruyne. A worldwide flock of Condors : Load sharing among workstation clusters. Journal on Future Generations of Computer Systems, 12, 1996.

No context found.

J. Pruyne. A Worldwide Flock of Condors: Load Sharing among Workstation Clusters. FGCS, vol. 12, 1996, 53#66.

No context found.

D.H.J. Epema, M. Livny, R. van Dantzig, X. Evers, and J. Pryne. A worldwide flock of condors: Load sharing among workstation clusters. http://www.cs.wisc.edu/ condor/doc/flock.ps, 1995.

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