Measure locally, explain locally Craig A. Stewart Citation: Newby, G., Apon, A., Berente, N., Eigenmann, R., Fratkin, S. Lifka, D., & Stewart, C.A. (2014, Novemeber). Return on investment from academic supercomputing: SC14 Panel. Presented at SC14 – the international conference for high performance computing, networking, storage and analysis. New Orleans, LA. Retrieved from http://hdl.handle.net/2022/21676 ___

Core Hours delivered by HPC systems

Support grant dollars vs. HPC investments

Statistics Gathering Context Tap into operational statistics of every group Batch system, file system, tape archive,… Tie into enterprise information systems Identity management, financial systems Only store derived information Relate scientific value and IU grant successes to investment in HPC operations (people and software) as well as HPC Hardware

Technology MySQL database Ruby on Rails scripting CAS authentication Connection to data sources Cron jobs SQL connections File based importing Plan: Leverage XDMOD

Initial observations Strong correlation that well funded researchers make extensive use of IU HPC systems F&A formulae are problematic: we are always over the limit for the category that includes cyberinfrastructure BUT…. Initial results suggest that F&A return on grants to users of HPC system are several times larger than the cost of purchasing and operating our HPC systems In the end we need to be able to explain the financial benefits to the financial people, and the scientific benefit to scientists.