Research Computing is a critical component for many science, social science and other research areas. Scientists and researchers are constantly pushing at the limits of what computation can offer as they discover new forms of matter, develop deeper understanding of genomics and biology, predict weather patterns, gain knowledge on the workings of the universe and much more. For many of these efforts, part of the challenge is developing new computational methods and models to support the kinds of science that is being attempted. These new approaches need to be validated, tested, and pushed to scale as part of advancing new science. Additionally, existing models and approaches continue to leverage the finite and limiting infrastructure that exists.
Often researchers scale back their science to fit within the constraints of their environments, while other researchers are limited to existing approaches because centers are not designed to easily add new software systems. With all these demands on the computing infrastructure (the need to validate new approaches, the need to support existing efforts, the increasing number and problem size for many disciplines), access to the right type of computing resources, at the right time, in the right quantity is a huge challenge.
Cycle Computing software combined with cloud-based compute and storage enables a flexible set of resources that can scale to meet the problem size, easily spin up and test new approaches, and provide near instant access to new resources when needed – simply, efficiently, and effectively. Cycle Computing’s CycleCloud software suite is the leading cloud orchestration, provisioning, and data management platform for science and research applications running on any cloud or internal environment.
NASA Research Example:
A project within NASA was working to determine the biological mass of bushes and trees over the entire arid and semi-arid zone on the south side of the Sahara. As part of a proposal to take part in the Head in the Clouds Program with AWS, Cycle Computing, and Intel Corp., Compton Tucker at NASA Goddard Space Flight Center and Paul Morin at the University of Minnesota developed plans to estimate the carbon stored and establish a baseline for later research on the expected carbon dioxide uptake on the south side of the Sahara.
With the recent access to large amounts of high-resolution remote sensing data, it is now possible to measure biomass at the level of individual trees and bushes. The team were initially using the Advanced Data Analytics Platform (ADAPT)—a private cloud within the NASA Center for Climate Simulation (NCCS). Like many private clouds, ADAPT is shared with a large number of other research projects, and the NASA team concluded additional computing power and storage capabilities were necessary to successfully conduct this research in a meaningful time frame.
By partnering with Cycle Computing and leveraging the CycleCloud software with AWS cloud resources, NASA was able to obtain results faster than previously possible and in a secure and cost-competitive way.
To do this, the team developed a workflow within CycleCloud that enabled them to capture satellite data, store it locally and do an initial pre-process to remove non-relevant images. CycleCloud then started moving data into the cloud. Once a relevant amount of data was available, CycleCloud started the cluster to begin processing the data while additional data was still being delivered. This allowed for very efficient use of resources, while getting results as fast as possible. Once the data analysis was complete, CycleCloud returned the relevant data to the local storage and cleanly closed down the cloud cluster.
The group is well on its way toward measuring the first study domain across Saharan Africa using a modest 8 terabytes of data and leveraging 200 virtual machines. These initial runs have proven the scalability of the cloud environment and its ability to accelerate the time to results and discovery. The initial scalability runs proved that all 11 study domains, approximately 100 terabytes of data, could be completed in the same amount of time it took for a single domain at a very competitive cost point.
For more information on this project, visit the NASA@SC15 website.
Beyond the above effort, Cycle Computing has been helping researchers a wide range of disciplines and problem sizes.
Learn more about Cycle Computing’s CycleCloud software suite