A Social Cloud for Public eResearch

23
A Social Cloud for Public eResearch Koshy John Kris Bubendorfer Kyle Chard School of Engineering and Computer Science Victoria University of Wellington Wellington, New Zealand School of Engineering and Computer Science Victoria University of Wellington Wellington, New Zealand Computation Institute, University of Chicago and Argonne National Laboratory Chicago, IL, USA

description

Scientific researchers faced with extremely large computations or the requirement of storing vast quantities of data have come to rely on distributed computational models like cloud computing. However, distributed computation is typically complex and expensive. The Social Cloud for Public eResearch aims to provide researchers with a platform to exploit social networks to reach out to users who would otherwise be unlikely to donate computational time for scientific and other research oriented projects. In this paper we explore the motivations of users to contribute computational time and examine the various ways these motivations can be catered to through established social networks. We specifically look at integrating Facebook and BOINC, and discuss the architecture of the functional system and the novel social engineering algorithms that power it.

Transcript of A Social Cloud for Public eResearch

Page 1: A Social Cloud for Public eResearch

A Social Cloud for Public eResearch

Koshy John Kris Bubendorfer Kyle ChardSchool of Engineering and

Computer ScienceVictoria University of Wellington

Wellington, New Zealand

School of Engineering and Computer Science

Victoria University of WellingtonWellington, New Zealand

Computation Institute,University of Chicago and

Argonne National LaboratoryChicago, IL, USA

Page 2: A Social Cloud for Public eResearch

Public eResearch

• A cure for cancer? AIDS?• Better cryptographic systems?• Investigating climate change?• Searching for intelligent life out there?• What do they have in common?

Page 3: A Social Cloud for Public eResearch

eResearch with BOINC

• Berkeley Open Infrastructure for Network Computing• Using BOINC for inexpensive computation• Unused resources for worthy causes• Making science accessible• But is it achieving its potential?

Page 4: A Social Cloud for Public eResearch

Facebook Statistics

• On average, people on Facebook install apps more than 20 million times every day

• Every month, more than 500 million people use an app on Facebook or experience Facebook Platform on other websites

• More than 7 million apps and websites are integrated with Facebook

BOINC Statistics• BOINC projects have just 2.3 million users in total• Only ~0.5 million are actively contributing

https://www.facebook.com/press/info.php?statistics

Page 5: A Social Cloud for Public eResearch

But…• BOINC marshals 5.5 petaflops from 0.5 million

Imagine if…

• Even 1% of Facebook’s user base joined & contributed actively• That’s 8 million users vs. the current 0.5 million• 90 petaflops?? – ballpark figure

At 10% and with hardware advances, we hit EXASCALE computing.A viable solution to what Brian P. Schmidt asked of eScience??

Page 6: A Social Cloud for Public eResearch

What’s holding BOINC back?

• Lack of visibility• Barriers to entry for new users• Lack of ease in identifying and joining new projects• Low visibility for new projects• Low levels of active contribution

Page 7: A Social Cloud for Public eResearch

A Social Cloud

“A Social Cloud is a resource and service sharing framework utilizing relationships and policies established between members of a social network.” – Kris Bubendorfer

Page 8: A Social Cloud for Public eResearch

A Social Cloud for Public eResearch• A Facebook application• Increasing awareness• Making use of social relationships• Truly bringing science to the people• Potential for growth – going viral

• How is this better than just BOINC? 800million

Page 9: A Social Cloud for Public eResearch

Deep Facebook Integration

• The Graph API – read/write• The OpenGraph Protocol – entities as objects• Social Channels– News feed– Requests– Notifications

Page 10: A Social Cloud for Public eResearch

Social Engineering (part I)

• Easing the Process of Joining– Interest Signature– Project Signature– Signature Distances (resource share)

Page 11: A Social Cloud for Public eResearch

Social Engineering (part II)• Incentivising Involvement, Contribution & Growth– Project Champions• Total project credits• Easier to become one of a less popular project• Encourages contribution• Champions the cause of a project – helps new users

Page 12: A Social Cloud for Public eResearch

Social Engineering (part II)• Incentivising Involvement, Contribution & Growth– Social Anchors• Social Values and Social Scores

• To motivate users to encourage their friends to join• Top bracket local to user’s social cloud

Page 13: A Social Cloud for Public eResearch

Social Engineering (part II)• Incentivising Involvement, Contribution & Growth– Compute Magnates• Rolling credit value, compute value, compute score

• Social pressure on friends to contribute• Top bracket local to user’s social cloud

Page 14: A Social Cloud for Public eResearch

Architecture

Page 15: A Social Cloud for Public eResearch

InteractionsSimple example of interactions between all actors associated with the Social Cloud for Public eResearch.

Page 16: A Social Cloud for Public eResearch

The Prototype

Page 17: A Social Cloud for Public eResearch

Results

• Verified interest signatures, project signatures and signature distances through a user study.

• Studied the effects of project champions, social anchors and compute magnates through simulations on the IEEE VAST 2009 Challenge dataset.

Page 18: A Social Cloud for Public eResearch

Results – Signatures (small sample)

Page 19: A Social Cloud for Public eResearch

Results – Project Champions

Page 20: A Social Cloud for Public eResearch

Results – Social Anchors

Page 21: A Social Cloud for Public eResearch

Results – Compute Magnates

Page 22: A Social Cloud for Public eResearch

How is the Social Cloud better?

• GridRepublic• Intel’s Progress Thru Processors

Page 23: A Social Cloud for Public eResearch

Conclusion• Increased visibility and engagement• More computational power for researchers• Science becomes meaningful to the masses• Brings lesser known projects into the limelight

• Science benefits. Humanity benefits.