NERC in Use with Jonathan Appavoo
We sat down with Jonathan Appavoo, an Associate Professor of Computer Science at Boston University, to discuss how he utilizes the New England Research Cloud (NERC) to elevate his research. The NERC is a common cloud framework with world-class cloud computing services and is a core project of the MOC Alliance. The NERC provides a suite of services (IaaS, PaaS, SaaS) that are not readily available today. To learn more about the core services and applications of the NERC visit the MOC Alliance NERC Tutorials page.
Appavoo is leading and contributing to multiple projects funded through the Red Hat Collaboratory that utilize NERC resources. One of Appavoo’s research thrusts involves recording and analyzing the steps a computer takes when executing a program to train the system to learn from its past patterns and improve its performance. As a stepping stone to this goal, Appavoo has constructed a computer simulator that carefully records its entire state after every low-level instruction it executes. Running programs in this simulator—such as an early computer chess game—allows him to generate large rich data sets for him and his team to study. The goal of this research is to explore the possibility of augmenting computer systems with biologically inspired mechanisms that allow the systems to automatically improve their efficiency by detecting and exploiting redundancy in their behavior.
Appavoo described how his research requires vast amounts of data to capture all of the computer’s steps. His research also requires an extensive platform that can generate these large data sets, store the data, and process it. He also desired a way to share simplified versions of the data produced by his team of systems researchers with data scientists at Red Hat to analyze. With these specific technical requirements, Appavoo needed a comprehensive platform that could be quickly configured to his research needs and propel his research forward stating “Building the tools is not good enough, you have to run it as well, and we need storage for all of the data.”
The NERC was able to not only meet all the requirements to conduct Appavoo’s research but also allowed him to configure the NERC’s interface to create and customize the processes to execute each step of his research. The NERC provided the underlying infrastructure needed for the automation Appavoo’s research required. It allowed him to quickly configure Virtual Machines (VMs) with all the necessary software to develop and run his unique and custom simulation software as well as the necessary standard data analytics and machine learning tools he requires to process the data his simulator produces. His team was also able to easily configure storage volumes to stage and archive terabytes of compressed data. The flexibility in the resources offered by the NERC has allowed Appavoo to mix the use of VMs configured with and without GPUs to meet the differing needs of the data production and analysis stages of his work. Finally, the automation support of the NERC allowed Appavoo to set up a simple on-demand usage model for his VMs. In particular, the VMs automatically shut themselves down when no one on his team is actively using them to run simulations or process data. However, when anyone on his team attempts to access the VMs via “SSH”, a standard method for accessing remote computers, the VMs automatically start themselves back up. This way his costs are minimized and he is only charged for the time his team actively uses NERC resources.
Additionally, Appavoo praised the ease and convenience of the NERC. He described how quick the process was to fully customize the NERC interface to his needs and how the MOC Alliance and NERC teams were always available and efficient in providing solutions for any complications stating “With just a few clicks of a button you get the storage you need and can arrange it across the virtual machines.” With the NERC Appavoo was able to create a cost-effective alternative to public clouds, giving him more time and resources to cultivate his research.