Social Cloud: Cloud Computing in Social Networks

With the increasingly ubiquitous nature of Social networks and Cloud computing, users are starting to explore new ways to interact with, and exploit these developing paradigms. Social networks are used to reflect real world relationships that allow users to share information and form connections between one another, essentially creating dynamic Virtual Organizations. We propose leveraging the pre-established trust formed through friend relationships within a Social network to form a dynamic “Social Cloud”, enabling friends to share resources within the context of a Social network. We believe that combining trust relationships with suitable incentive mechanisms (through financial payments or bartering) could provide much more sustainable resource sharing mechanisms. This paper outlines our vision of, and experiences with, creating a Social Storage Cloud, looking specifically at possible market mechanisms that could be used to create a dynamic Cloud infrastructure in a Social network environment.

1. Social Cloud: Cloud Computing in Social Networks Kyle Chard, Simon Caton , Omer Rana and Kris Bubendorfer

2. Emerging Themes Cloud Computing is growing in strength Many providers e.g. Amazon EC2/S3 , Google App Engine, Microsoft Azure and also many smaller scale open clouds such as Nimbus and Eucalyptus. Social Networking is increasingly ubiquitous: E.g. Facebook has over 400 Million active users. 50 % of these users log on every day

3. Current Cloud Scenarios and Problems Sharing Finite capacity vs. fluctuating requirements Many social peers with different capabilities Economy Small scale consumers have ad hoc requirements Money grabbing providers and inflexible lock-in Trust always assumed at some level Anonymity (Market-based/broker allocation) Many models fall apart when this is removed

4. Social Networks Formed through pre-existing relationships, i.e. your friends Have a pre-existent fabric of trust inherently interwoven into the network How many of your friends do you not trust? Many applications now use social networks as a platform for: Authentication e.g. Facebook Connect Application Portals e.g. ASPEN and PolarGrid projects There already exist well established application APIs

5. The Social Cloud Vision + + An amalgamation of: Social Networking Cloud Computing Volunteer Computing A Social Cloud allows friends to share capabilities within the context of a Social Network. Volunteer computing arises as users can share resources for little or no gain , perhaps through reciprocal arrangements . The leveraging of pre-existing relationships in order to enable mutually beneficial interactions within a cloud context.

6. Social Cloud Interaction Vision Social Cloud Socially – orientated Market Place

7. Social Cloud Economy Payment (in an economic sense) is optional Instead we utilise a virtual currency All collaborations involve a transfer of “credits” All participants are given an initial amount of credits No one can buy additional credits – they must be earned Therefore, we can prevent free-riding, and actively encourage participation

8. Community Effect Susceptible to cheating through fabricated accounts Social Enforcement: exclusion of anti-social peers To encapsulate the nature of an interaction an agreement is used for the domains: Technical Requirements Non-functional properties Temporal Requirements Economic preferences WS-Agreement + EJSDL + DRIVE API + Reservation + Social Cloud Extensions

9. Social Cloud Proof of Concept Simple Storage Service Implemented as a Facebook application Use Case: a back up facility Agreement

10. Posted Price Storage Social Cloud Storage Storage MDS User ID URL Capacity Price User1 100MB 5 User2 500MB 10 User3 5GB 7 Enables interactions based upon active trading/collaborative decisions Intuitively facilitates reciprocal collaboration Current “norm” in industry solutions

11. Dynamic Auctions Auction: Enables dynamic participant pairing Sealed bid second price reverse auction Could be extended to any other auction mechanism

12. Evaluation Research Questions: Can a Social Cloud Scale? What are the computational requirements for an “average” sized Social Cloud? According to Facebook , the average social network has 130 participants Can a Social Cloud function in a timely manner as a Facebook application?

13. Posted Price Scalability Varying the size of the MDS and number of matches With a size of 2000, 100 matches can be discovered in ~ 2 seconds, which is reasonable

14. Auction Scalability 500 Auctions and the worst case scenario: all auctions run concurrently Even with 50 bidders can still complete 65 auctions per minute Under our assumptions this is already enough for a large social network

15. Dissemination of Results A social (storage) cloud can be hosted using minimal resources (3 – 4 yr old PC) Components show good throughput under realistic loads However, scaling to millions of users would require a dedicated HPC or elastic environment Co-op model  members sustain the platform

16. Conclusions & Future Work Social Cloud Dynamic cloud environment leveraging existing trust relationships Proof-of-concept: can be extended for many new scenarios Future Work Computation, licenses and other capabilities Combinatorial auctions Generic scientific cloud communities – e.g. myExperiment Evolution of the economic model

17. Questions? Please look at our Prototype Social Cloud Video http://www.im.uni-karlsruhe.de/SocialCloudDemo [email protected] / [email protected] [email protected] [email protected] [email protected]