Spring 2003 Workshop on Visualization Frameworks

Table of Contents

Date and Location

April 14-April 15, 2003 at the Woodfin Suites, Emeryville CA Link to Woodfin Suites website.

There is a block of eight rooms at the Woodfin Suites is under the group name Visualization Frameworks Workshop. Dates blocked are from Sunday, April 13 to Wednesday, April 16. There should be no problem extending nights on either end. Cost of rooms is $119 +12% tax per night. Cut-off date for reservations is April 1, at which point prevailing rates will apply. (N.B. - the workshop will conclude at about 4:30pm on Tuesday April 15. Staying over the night of the 15th is your choice, but there will be no further workshop activities after approximately 4:30pm on 4/15.)


Background and Workshop Objectives

Scientific visualization - the transformation of abstract information into readily comprehensible images - plays an integral part of the scientific process. Over the years, scientific visualization has evolved to keep pace with advances in computer science, software engineering, data modeling and data management. The earliest scientific visualization "systems" were subroutine APIs, and each new visualization application required writing software to read data and perform subroutine calls to perform visualization and rendering. During the 1990s, a class of visualziation systems emerged that were highly successful. These applications were composed of visualization software components, an interface for raplidy constructing an application using the provided components, or possibly new components created by a developer. An internal "flow executive" scheduled execution of these components and moved data through the network of software components. These systems were unexpectedly successful because they were extensible, since developers could extend the functionality of the base system through the addition of custom software components. Communities of users and developers coalesced around each of these systems, for a common framework and data model fostered a stable environment for sharing components, data and ideas.

These dataflow systems, while immensely successful, have inherent design limitations that prevent their use, or adaptation for use, in a class of contemporary and near-term environments. Remote and distributed visualization environments pose new challenges not anticipated by the designers of flow-based visaualization technologies of the 1990s. The challenges and shortcomings are reflected in the list of technical topics to be addressed by the workshop (below). Over the years, many projects have been undertaken over the years to adapt flow-based visualization systems for use in RDV environments, yet none have been successful. As a result, there has not emerged a community concensus concerning a framework for high performance, remote and distributed scientific visualization (RDV). RDV applications of today are best characterized as prototype research projects that excel in one or two aspects of the end-to-end visualization task.

The purpose of the proposed workshop is to bring together a group of visualization experts to tackle the problem of defining the scope and architecture of an effective visualization framework suitable for use in an RDV setting. The experts represent a cross-section of authorities from scientific visualization, with emphasis upon software engineering and architecture, and with first-hand experience designing and building visualization software. The workshop results will consist of a technical report that defines research, development and deployment challenges to be met in order to realize an RDV framework. Because of the breadth and depth of technical topics pertaining to a unified RDV framework, this workshop is viewed as a "groundbreaking" activity that will lead to more discussion and topical refinement in future meetings.

Ultimately, it is our hope that this meeting will be the first in an ongoing process that will result in a framework suitable for use by a wide community of visualization researchers and scientists. One hindrance to visualization research is the lack of a common data model, and lack of a stable and consistent testbed, particularly for work on RDV algorithms and infrastructure. Such a framework will provide a stable deployment foundation so that the fruits of visualization research can find their way into the hands of the scientists, who are in need of such tools. An open framework creates the possibility of greater unification within the community of visualization researchers and scientists alike, for it takes real steps towards meeting the needs of both groups.

The workshop format will consist primarily of discussion forums. Each forum will focus on a narrow range of technical topics, with the objective being a narrative describing challenges and objectives, along with a list of milestones along the path towards realizing a fully operational RDV framework that can be used by a large community of visualization and computational scientists.

The anticipated list of topics (discussion focus areas) is as follows (subject to change):

  • Toolset vs. Framework? Or, what is the right matrix of technologies to define the scope of an RDV infrastructure (vertical, horizontal).
  • Data management/modeling requirements for component-based visualization software tools.
  • Grid technologies that might be useful for an RDV framework, or Grid technologies that are needed but missing.
  • Network transport of data amongst distributed components.
  • Component architectures and interfaces. Data transport and data delivery adaptors.
  • Realizing a framework. Does this eventually show up on Sourceforge?
  • Development and research roadmap.

Pre-Workshop Homework

Due 1 April 2003

Please provide a 500 to 1000 word position statement no later than 1 April 2003. The position statement should be a collection of thoughts on the subject of visualization framework requirements. I.e., what are the primary requirements you envision for a framework as described in the workshop objectives. Feel free to include references to items you feel pose substantial technical or social challenges. Please email your position statement directly to Wes Bethel (ewbethel@lbl.gov) no later than 1 April 2003. Use whatever format you prefer, so long as your name is included. These statements will be used to guide the workshop agenda, and will not be made public except to other workshop participants.




Agenda (in progress)

Monday 4/14/2003

8:30am - 5pm - workshop
6:30pm - 8:30pm - Dinner in Berkeley on Telegraph Avenue at Berkel Berkel

Tuesday 4/15/2003

8:30am - 4:30pm - workshop

For your planning purposes, we will conclude all workshop activities by 4:30pm on Tuesday 4/15.


Attendees

John van Rosendale, DOE-HQ
Wes Bethel, LBL 
John Shalf, LBL
Randy Frank, LLNL
Dean Williams, LLNL
John Clyne, NCAR
Jim Kohl, ORNL
Steve Parker, Utah
Joel Welling, PSC
Pat Moran, NASA-Ames
Ron Kriz, Virginia Tech
Ken Joy, UC Davis
Sam Fulcomer, Brown 

Workshop Findings

The workshop findings are presented in the following PDF file. RDVWorkshop-Findings-Document-051403.pdf