Science Cloud Summer School

SALSA HPC Group, Indiana University


Prerequisites: Assumes experience in scientific computing including Linux, MPI and Service oriented architectures

Introduction: The Science Cloud Summer School (SCSS for short) is intended to be a 5 day event, with physical participation of attendees as well as virtual participation (synchronous and/or asynchronous). Indiana University is committed to providing the organization lead and facilities for hosting the physical meeting and supporting virtual participation - with the tentative site being IU's facilities in Indianapolis. The will leverage expertise in various aspects of cloud computing from FutureGrid; naturally, FutureGrid personnel will be recruited to provide content to the school, but in addition we will seek participation of experts in the field outside the project who will provide guest lectures. A key component of the school will be actual hands-on experimentation with cloud technologies – here, FutureGrid resources will also be leveraged, providing students with access to a rich, diverse set of middleware for clouds (Nimbus, OpenStack, Hadoop, etc). In addition, we will seek partnerships with commercial providers (including Amazon and Microsoft) to seek their donation of “credits” to allow hands-on experiments on commercial cloud infrastructures (this is common practice in Amazon EC2 for education, and we have active relationship with Microsoft Azure as well). We will market the event as one with significant hands-on activities on research and commercial clouds, and we anticipate that this will pique the interest of graduate students and their advisors.

The SCSS would be branded as an event that is part of the larger XSEDE scope. We anticipate there will be considerable interest from the XSEDE community, and with support from XSEDE in endorsing and performing marketing/outreach for this event (e.g. through campus champions, mailing lists, etc), we anticipate that SCSS can target of the order of ~100 attendees. To cover expenses for physical attendance by a significant fraction of the students, the intent is to seek funding from the NSF for participant costs.

Remote participation will be provided with resources already available at IU, with collaboration technologies that favor ease of accessibility (e.g. Adobe Connect, Webex). The detailed approach for how SCSS will support virtual participation is still being fleshed out, but overall we plan to have synchronous “webcasting” of presentations, self-paced educational modules remote users will be able to follow on their own, and the use of social media and collaboration tools (Google+/hangouts, twitter, FaceBook) to keep remote participants engaged. We also plan to promote a competition/prize, and offer the opportunity of participants to submit a paper to the “Concurrency: practice and experience” journal on the broad area of cloud computing and applications in science to drive students. This will be modified as part of VSCSE

If this event is as successful as we expect it to be, the goal is for the SCSS to be a recurring event – we anticipate for at least the next three years interest will be quite strong for such an event.

Future Goals: We can explore holding it in Europe and China; We can also use to invigorate our FutureGrid community groups (MapReduce and science Clouds) and offer student prizes

References
Big Data for Science Workshop at NCSA virtual summer school (VSCSE 2010), July 26-30, 2010
Cloud Technologies, Data Intensive Science and TeraGrid, Tutorial for (TeraGrid 2010 ) Conference, August 2-5, 2010