Ribbons: Enabling the Effective Use of HPC Utilization Data for System Support Staff

Loading...
Thumbnail Image
Date
2014
Journal Title
Journal ISSN
Volume Title
Publisher
The Eurographics Association
Abstract
Beyond raw computational power, a supercomputer offers the capability of generating and logging a significant amount of diagnostic data. While adding to the burden of maintenance, this data nevertheless represents compelling opportunities for development directed toward improved evaluations, diagnostics, analytics, etc. We have developed such a utility, a visual analytics tool for the support staff of the Blue Waters supercomputer. Our initial goal was broad: provide an informative illustration of current running jobs on the machine for the purpose of system monitoring. Additionally, we were able to collect diverse utilization data to the extent that both minimizing exclusion of as well as intuitively coordinating information were equally challenging. Our primary visual element is an extension of a stacked bar chart to increase horizontal continuity; resulting visualizations show system utilization as a series of concurrent job ''ribbons''. The remaining elements are common visual/interactive techniques offering expansive functionality. Together these components were deployed as a web application, which is referred to as the ''ribbon viewer'' by its regular users. In this paper we will highlight the design nuances and development complexities that are belied by the ribbon viewer's apparent simplicity. We will also discuss use-case scenarios in terms of both typical usage and specific examples.
Description

        
@inproceedings{
:10.2312/eurova.20141139
, booktitle = {
EuroVis Workshop on Visual Analytics
}, editor = {
M. Pohl and J. Roberts
}, title = {{
Ribbons: Enabling the Effective Use of HPC Utilization Data for System Support Staff
}}, author = {
Sisneros, Robert
and
Fullop, Joshi
and
Semeraro, B. David
and
Bauer, Greg
}, year = {
2014
}, publisher = {
The Eurographics Association
}, ISBN = {
978-3-905674-68-2
}, DOI = {
/10.2312/eurova.20141139
} }
Citation
Collections