Supporting Distributed Visualization Services for High Performance Science and Engineering Applications A Service Provider Perspective

The eVis system is being used with users undertaking atmospheric and oceanography based climate model analysis, typically using five dimensional data sets with a range of parameters with each data set being about 200 MB. It is used for studying molecular dynamics, vortex shredding around aircraft wings and similar. Initial experience show that the users require intense one to one support to navigate through the first step of acquiring a digital certificate and getting to register with the system to all the way through the compute clusters to the visualization node, display and render services, the use of the visualization toolkits with virtualGL. In addition, the users also require technical help to read in their output data from simulations into the visualization toolkits, introduction to new techniques available through these systems. The end results so far have been positive but the support required per user is about a person over a period of three months, more if the user is a complete novice. Setting up and running these systems are equally labour intensive. A full production service should ensure that not just hardware and administrative staff but a pool of visualization support service personnel are also in place.