Citation
Daniel Zinn. "Parallel Virtual Machines in Kepler". Talk or presentation, 16, April, 2009; Presented at the 8th Biennial Ptolemy Miniconference.

Abstract
Scientific Workflow Systems provide an intuitive and easy interface for users to design applications for analysis and processing of data. They greatly simplify the understanding and sharing of complex computation processes and do not require the user to have any prior programming experience. An often-used model of computation for Scientific Workflows is "Process Networks" (PN), in which concurrently running processes communicate with each other via pre-defined channels. We developed a light-weight parallel PN engine, called PWS, that allows to execute PN workflows in a cluster environment. We furthermore extended Kepler to function as a front-end to PWS. The users are thus able to use the intuitive Kepler GUI (an adoption of Vergil, the Ptolemy II GUI) to build and execute PN workflows that can run locally as well as in a cluster environment utilizing parallel resources. Our demonstration will show how a movie transformation pipeline can be built within Kepler and then be executed locally as well as on a cluster computer. Kepler is loosely coupled to the cluster PN and thus the data processed on the cluster is not routed through Kepler. Instead, Kepler is used only to setup, launch and monitor cluster PN workflows. However, by exploiting the communication channel from Kepler to PWS, Kepler actors and PWS actors, coexisting in the same workflow, can communicate with each other through String ports.

Electronic downloads

• 04-Zinn_ParallelVirtualMachinesInKepler_Apr09.pdf · application/pdf · 18337 kbytes

• HTML

  Daniel Zinn. <a
  href="http://chess.eecs.berkeley.edu/pubs/553.html"
  ><i>Parallel Virtual Machines in
  Kepler</i></a>, Talk or presentation,  16,
  April, 2009; Presented at the 8th Biennial Ptolemy
  Miniconference.

• Plain text

  Daniel Zinn. "Parallel Virtual Machines in
  Kepler". Talk or presentation,  16, April, 2009;
  Presented at the 8th Biennial Ptolemy Miniconference.

• BibTeX

  @presentation{Zinn09_ParallelVirtualMachinesInKepler,
      author = {Daniel Zinn},
      title = {Parallel Virtual Machines in Kepler},
      day = {16},
      month = {April},
      year = {2009},
      note = {Presented at the 8th Biennial Ptolemy
                Miniconference},
      abstract = {Scientific Workflow Systems provide an intuitive
                and easy interface for users to design
                applications for analysis and processing of data.
                They greatly simplify the understanding and
                sharing of complex computation processes and do
                not require the user to have any prior programming
                experience. An often-used model of computation for
                Scientific Workflows is "Process Networks" (PN),
                in which concurrently running processes
                communicate with each other via pre-defined
                channels. We developed a light-weight parallel PN
                engine, called PWS, that allows to execute PN
                workflows in a cluster environment. We furthermore
                extended Kepler to function as a front-end to PWS.
                The users are thus able to use the intuitive
                Kepler GUI (an adoption of Vergil, the Ptolemy II
                GUI) to build and execute PN workflows that can
                run locally as well as in a cluster environment
                utilizing parallel resources. Our demonstration
                will show how a movie transformation pipeline can
                be built within Kepler and then be executed
                locally as well as on a cluster computer. Kepler
                is loosely coupled to the cluster PN and thus the
                data processed on the cluster is not routed
                through Kepler. Instead, Kepler is used only to
                setup, launch and monitor cluster PN workflows.
                However, by exploiting the communication channel
                from Kepler to PWS, Kepler actors and PWS actors,
                coexisting in the same workflow, can communicate
                with each other through String ports. },
      URL = {http://chess.eecs.berkeley.edu/pubs/553.html}
  }
  

Posted by Christopher Brooks on 17 Apr 2009.
Groups: ptolemy
For additional information, see the Publications FAQ or contact webmaster at chess eecs berkeley edu.