Academic and scientific research often involves the construction of mathematical and numerical models to solve scientific and engineering problems. Traditionally, these complex and intensive computational models have been implemented on super computers or high-performance computing (HPC) infrastructure. These models are difficult to setup and operate, and can create a painful experience for researchers who often have to wait in a long line to use their university's super computing infrastructure, whether it's for a few hours or a few days.
Although universities have used cloud-based applications for years (e.g., email), today the cloud computing trend is quickly evolving into a premium model for scientific computation and Big Data. However, the idea of running HPC in public clouds has both limits and specific use cases today.
When it comes to scientific computations and numerical models, we can categorize the workloads into the following:
- Workloads with a high degree of parallelism that can ideally run in public clouds.
- Workloads with parallelism but with a high degree of dependencies between nodes. These workloads are extremely chatty and require very fast communication between compute nodes (high throughput network bandwidth and latency).
The second category describes the majority of HPC environments, and many public clouds are not ready or optimized for these workloads. However, an OpenStack-powered open cloud provides researchers a completely new model to utilize hybrid computing infrastructure (private cloud for on-campus, public cloud for off-campus). A hybrid model makes it more viable for researchers to rent high performance computing.
Why Open Cloud for Universities?
Universities have always been at the forefront of innovation because of their focus on research and their embracing of open standards and collaboration. When hundreds of researchers contribute to a shared purpose, everyone benefits. In fact, open-sourced software has proven that proprietary ownership often precludes innovation.
The scientific community can trust the wisdom of collaborative communities with common goals for following reasons:
- An open cloud fosters collaboration. When researchers use an open cloud to build their scientific research, they know their experiments run anywhere in the world.
- An open private cloud based on OpenStack provides researchers with the flexibility they need to build customized high bandwidth and high performance clouds with full interoperability and compatibility with OpenStack-based public clouds.
- For years, the fundamental mission of universities has been teaching and research. Open source technologies empower scientists to join, collaborate with and contribute to open source communities, such as Open Compute and OpenStack. They can leave lasting changes to the computing industry.
It’s amazing to imagine that Rackspace, with its humble idea of open sourcing the cloud, is enabling collaboration and the wisdom of the communities to help shape academic and scientific researchers push the envelope and forge the next generation of innovation.