Data Storage Meet-up: Challenges and Emerging Architectures for Effective, High-Performance Research Data Storage
TimeWednesday, July 2510am - 11am
DescriptionModern scientific computational workloads are putting increasing strain on existing storage architectures. A decade or more ago, scientific applications were typically CPU bound. Today, many computational workflows such as those in the life sciences and data analytics are heavily IO bound and frequently involve large input, output, and temporary data sets and lots of random IO.
Traditional storage systems simply weren't designed for the scalability and performance requirements of modern scientific workloads. The emerging use of machine learning, GPUs, and other accelerator technologies make the situation even more challenging. New storage architectures and solutions are needed to ensure that data storage will be able to keep up with the future computational demand.
Join us for three short presentations to set the stage for lively open discussion and networking. Topics include some of the key storage challenges in computational research, emerging software solutions that have shown great promise, and storage characteristics needed to support state-of-the-art scientific research.
Overview of Storage Challenges in Scientific Research
Markus Dittrich, BioTeam
Evaluating Emerging Flash Storage Architectures for Research Computing
Ron Hawkins, SDSC
Bending the Rules of Reality for Improved Collaboration and Faster Data Access
David Hiatt, WekaIO