HUBzero®: Novel Concepts Applied to Established Computing Infrastructures to Address Communities' Needs

The science gateway framework HUBzero® has been enhanced and further developed since its initial vision in 1996 -- always driven by requirements of the diverse communities applying HUBzero® for their research. HUBzero® is part of a computational landscape that has never evolved as fast as in the last decade. Novel frameworks and concepts on the user interface side such as Javascript libraries and Jupyter notebooks support communities in their working environment with easy-to-use user interfaces while novel technologies and concepts in the backend allow for effective and efficient modeling, simulations and processing research tools and data. HUBzero®'s enhancements include extensions for BOINC and XSEDE infrastructures on the backend while offering interactive computations and analytical tools via Jupyter Notebooks, RStudio, and other web applications as publishing environments. The paper goes into detail for novel developments for the three use cases Purdue University Research Repository (PURR), nanoHUB and MyGeoHub. First, PURR has been extended to utilize the enhanced data storage service Data Depot and high-speed networks of the Purdue local campus infrastructure. Second, nanoHUB offers over 500 simulation tools and it has been enhanced with a novel caching system to explore the input parameter space for already computed results via BOINC. The third extension is concerned with builtin features for geospatial data and modeling in MyGeoHub that offers to execute compute-intensive tasks on XSEDE. The diverse extensions can be reused in various hubs developed with HUBzero® requiring such diverse features and accessing different distributed computing infrastructures.