Wednesday, October 6, 2010

The Harvard High Performance Computing Summit

This is as close as I get to advertising on my blog - my group at Harvard Medical School (HMS) runs a yearly high performance computing summit to foster information exchange among colleagues.  Here's the story:

Over the past 5 years, the dramatic increase in biomedical data has catalyzed a need for high performance storage, high performance compute clouds, and new approaches to making information technology accessible to researchers.  At HMS, we have made great progress in addressing these needs and we continue to plan for the future.

The availability of genomic data and new information-rich data collection approaches of all kinds (such as imaging) require new approaches and new methodologies.  Computational methods, once a sideline for biomedical researchers, have become a required enabler of discovery.  Some laboratories no longer  have a wet-lab component, but instead use computational methods as their primary research method.  The scope and scale of research computing is no longer supportable by systems sequestered under the lab bench or in ad-hoc data closets.  The resources that are needed now require skilled IT professionals, scalable and fault tolerant infrastructure, and adaptive learning from other fields that have already undergone this transformation such as high energy physics. Storage requirements have grown from gigabytes to terabytes to the petabyte scale.   Processing power has grown from clusters of a dozen computers to thousands of simultaneous processing cores.

Unfortunately few of my colleagues in high performance computing have the time to share approaches and learn from their fellow leaders at other institutions. Four years ago we set out to solve this problem by creating the Biomedical High Performance Leadership Summit to exchange ideas, approaches and solutions to the challenges facing biomedical IT organizations. We are convening it again this year with the hope that the biomedical researchers and service providers can learn from other computational fields and build infrastructure to meet research needs.

This year's conference has four themes:

    * Scaling storage: managing storage in the age of the petabyte (and beyond)
    * Organizational scaling: building sustainable HPC centers
    * Collaborative computing: grids, clouds and open computing
    * Getting science done: making HPC work for researchers

I invite leaders in biomedical high performance computing to join me at the summit,  which includes keynotes from myself and George Church, industry updates from Chris Dagdigian and Addison Snell, and key lessons learned from John (Jay) R. Boisseau and Phil Papadopoulos.

For more information check out the website.

I hope to see you at Harvard for the Summit on October 17-19!

No comments: