Changes between Initial Version and Version 1 of ExptSvcsWGGEC3Minutes

11/10/08 10:58:10 (13 years ago)
Vic Thomas



  • ExptSvcsWGGEC3Minutes

    v1 v1  
     2        Minutes of the GENI Experimenter Workflow and Services
     3                        Working Group Meeting
     4                           30 October 2008
     6Recorded by Vic Thomas (GPO System Engineer)
     8The GENI Experiment Services and Workflow Working Group met as part of
     9the GEC-3 conference hosted by HP Labs, Palo Alto, California.
     11Meeting agenda:
     12    * Welcome and meeting objectives -- Jeff Chase, Duke University
     13    (WG chair)
     14    * Lifecycle of a GENI Experiment -- Vic Thomas, GPO (System
     15    engineer supporting the WG)
     16    * Project talk: GENI Experiment Control using GUSH-- Jeannie
     17    Albrecht, Williams College
     18    * Lightning talk: Slice Control and Programmability in Wireless
     19    Sensor Networks -- Hongwei Zhang, Ohio State University
     20    * Lightning talk: Orbit Management, Control & Measurement
     21    Framework -- Max Ott, NICTA
     22    * Project talk: Raven - A Provisioning Service for GENI -- John
     23    Hartman, University of Arizona
     24    * Project talk: Instrumentation Tools for a GENI Prototype -- Jim
     25    Griffioen, University of Kentucky
     26    * Lightning talk: Digital Object Architecture -- Giridhar
     27    Manepalli, CNRI
     28    * Project talk: Data Plane Measurements -- Deniz Gurkan,
     29    University of Houston
     30    * Next steps for the WG -- Vic Thomas, GPO
     32Jeff Chase described the scope and activities of the working group.
     33He showed where this WG fit into the overall GENI architecture and
     34introduced the talks to be presented at this meeting.  Jeff's
     35presentation includes a number of questions to be addressed by this
     36WG. The presentation also includes strawman GENI usage scenarios---
     37canonical usage scenarios that can be used for discussions on
     38experimenter workflow and services.  Time constraints prevented Jeff
     39from presenting this material; he hopes these issues will be discussed
     40on the WG mailing list and in future meetings.
     44Vic Thomas presented his view of the workflow for an experimenter
     45using GENI.  For each step of the workflow he identified tools and
     46services needed to support that step.  The objective of the
     47presentation was to get feedback on whether this workflow was
     48consistent with the thinking of the WG participants, to understand how
     49proposed projects and ideas fit into this workflow, and to identify
     50areas of disagreement.  A person in the room pointed out that it would
     51be useful to have a tool that schedules experiments based on resource
     52needs and availability.
     53Presentation URL:
     56Jeannie Albrecht described her project that is developing abstractions
     57and tools for addressing the challenges of managing multiple
     58applications.  The project is developing Gush, a distributed
     59experiment management infrastructure based on her previous work called
     60Plush.  Gush supports multiple steps in the experimenter's workflow:
     61Describing an experiment using application building blocks, acquiring
     62resources to run the experiment, configuring the resources, starting
     63the experiment, monitoring the experiment (including failure
     64handling), and finally cleaning up the resources used.  Gush is being
     65built to integrate with the PlanetLab control framework.
     66Presentation URL:
     69Hongwei Zhang gave a lightning talk on experimenter tools available on
     70Sensei, the wireless sensor network testbed from Ohio State.  This
     71testbed provides "slice control" services. These services allow slices
     72to be moved, i.e. the set of sensor nodes that belong to a slice can
     73be changed dynamically to track events of interest to the experiment.
     74They also help experimenters maintain slice invariants and collect slice
     76Presentation URL:
     79Max Ott gave a lightning talk on experiment related measurements on
     80Orbit.  The experimenter describes the experiment using a Ruby-based
     81language.  The information to be collected is also specified using
     82this language.  The measurements are collected in a relational
     83database.  Orbit provides APIs for analyzing the data collected.
     84Presentation URL:
     87John Hartman gave a project talk on Raven for managing software
     88deployments for GENI experiments.  Raven supports the deployment of
     89software to programmable components and updating this software while
     90the experiment runs.  It builds on the Stork project and like Stork is
     91targeted to the PlanetLab control framework.  Raven supports the
     92creation and distribution of packages and metadata files with configuration
     93instructions. Raven will also provide tools for resource discovery and
     94allocation.  Raven and Gush are complementary projects: Raven
     95addresses software management while Gush addresses slice management.
     96Presentation URL:
     99Jim Griffioen gave a project talk on the easy to use instrumentation
     100tools they are building so GENI experimenters have a better
     101understanding on the runtime behaviour of their experiments.  They are
     102modifying the Edulab web interface to include experiment monitoring
     103pages.  Examples of information that can be monitored include OS
     104performance/load, traffic graphs and packet traces.  Pictures and GUIs
     105are used to specify the information to be collected.  This project
     106complements Paul Barford's (U. of Wisconsin) project that is looking
     107at measurement data collection.
     108Presentation URL:
     111Giridhar Manepalli gave a lightening talk on CNRI's Digital Object
     112Architecture that includes a handle system, a digital object
     113repository and a digital object registry.  The handle system is a
     114distributed and secure policy management system.  It can potentially
     115be used for managing GENI principal records.  The registry and repository can
     116potentially provide registration, discovery and storage services to
     117GENI entities such as principal, component and slice registries, and
     118code artifacts.   Control frameworks interested in using the Digital
     119Object system should contact Larry Lannom or Giridhar Manepalli of CNRI.
     120Presentation URL:
     123Deniz Gurkan gave a project talk providing GENI experimenters with
     124measurements at the optical layer.  Examples of measurements include
     125bit error rate, optical power, and SONET and Gigabit Ethernet performance
     126monitoring.  This measurement work is especially useful for non-IP
     127networking experiments.  The project will identify feasible optical
     128layer measurement technologies  for GENI and will design a framework
     129that will allow experimenters to remotely configure and gather
     130instrumentation data.
     131Presentation URL:
     134Finally, Vic Thomas listed the formal deliverable of this working
     135group.  He, as the GPO system engineer, will draft these documents but
     136would appreciate active community involvement.  At the very least
     137community members should comment on the documents.  Text contributions
     138to the document will be greatly appreciated.  WG discussions will be
     139carried out on the WG mailing list.