Privacy Issues in Cloud computing Zeeshan Ali Shah System - PowerPoint PPT Presentation

Privacy Issues in Cloud computing Zeeshan Ali Shah System administrator PhD researcher KTH PDC Center for High Performance Computing W-Sunday, October 28, 12

Agenda • Zeeshan ¡Ali ¡Shah – Introduc5on ¡to ¡Cloud ¡compu5ng – Privacy ¡through ¡distribu5ng ¡data ¡ • Ali ¡Gholami – Privacy ¡in ¡Cloud ¡Compu5ng ¡Through ¡Iden5ty ¡ Management • QA 2 W-Sunday, October 28, 12

Big PaaS picture IaaS SaaS Elasticity Private TYPES Reliability Public Virtualisation Hybrid FEATURES MODES … … Local Cost ¡Reduction Cloud ¡ LOCALITY Remote BENEFITS Ease ¡of ¡use Systems Distributed … STAKEHOLDERS COMPARES ¡TO Service-­‑oriented Users Architecture Adopters Internet ¡of … Services Resellers Grid Providers 3 "Advances in Clouds" , Expert Group Report (2012), Editors Lutz Schubert, Keith Jeffery W-Sunday, October 28, 12

Deployment Hybrid models Private / Internal Cloud Public / External Public Cloud On Premises / Internal Off Premises / Third Party W-Sunday, October 28, 12

User View Cloud Clients (e.g. Browsers, Mobiles, Devices) Cloud Application (e.g. SaaS) Cloud Services (e.g. Web Services) Cloud Platform (e.g. App Server) Cloud Storage (e.g. Database) Cloud Infrastructure (e.g. Physical/Virtual Hardware) 5 Armbrust, M., Fox, M., Griffith, R., et al. (2009) “Above the Clouds: A Berkeley View of Cloud Computing”, In: University of California at Berkeley Technical Report no. UCB/EECS-2009-28, pp. 6-7, February 10, 2009 W-Sunday, October 28, 12

User View Cloud Clients Lock-in (e.g. Browsers, Mobiles, Devices) Cloud Application (e.g. SaaS) Cloud Services (e.g. Web Services) Cloud Platform (e.g. App Server) Cloud Storage (e.g. Database) Cloud Infrastructure (e.g. Physical/Virtual Hardware) 6 W-Sunday, October 28, 12

How to chose: The Cloud Mindmap SaaS user Want ready to go services Web applications PaaS Want high level tools to create my own services = SaaS provider User = IaaS user Want bare metal resources IaaS Utility computing W-Sunday, October 28, 12

Privacy issues, Why ? • Out sourcing of compute • Out sourcing of Data • Multi Tenant • Assurance for Cloud resource provider • Cross border regulations • .. W-Sunday, October 28, 12

Some ways .. • Encryption • Anonymization • see next W-Sunday, October 28, 12

Distribute the data .. • Keeping Data Private while Computing in the Cloud 2012 IEEE Fifth International Conference on Cloud Computing (see in supplement reading) W-Sunday, October 28, 12

The paper describes • Preserving data privacy while solving comput- intensive problems on un-trusted machines.. • With help of Data Distribution in chunks • It outperforms existing crypto-based privacy such as homomorphic encryption ---See results W-Sunday, October 28, 12

using tile assembly 0 0 0 0 1 1 1 1 0 1 1 0 1 0 0 1 0 0 0 0 0 1 1 1 0 0 1 0 1 1 1 1 0 1 0 1 0 1 0 1 (a) 0 1 0 1 0 0 1 0 1 0 0 1 0 1 0 0 1 0 1 0 0 1 0 1 0 0 0 0 1 1 0 0 1 1 0 0 1 1 0 1 0 1 0 0 0 0 0 1 1 0 0 1 1 0 0 0 0 0 1 0 1 1 0 1 0 1 1 1 0 0 1 1 0 0 1 1 1 1 0 1 0 1 1 0 1 0 1 1 0 1 0 1 1 (b) (c) (d) ` ` ` ` 1 2 1 2 ` ` 0 1 0 1 0 3 0 1 0 1 0 3 0 0 1 1 0 0 1 1 0 0 0 0 1 1 0 0 1 1 0 0 1 0 1 0 1 1 0 1 0 1 0 1 1 0 0 1 1 0 0 0 0 0 1 1 0 0 1 1 0 0 0 0 0 0 1 1 0 0 1 1 1 1 0 0 1 1 0 0 1 1 1 1 0 1 0 1 1 0 1 0 1 1 ` ` 4 4 ` ` ` ` ` ` ` ` 5 5 8 7 6 8 7 6 (e) (f) W-Sunday, October 28, 12

Overview of sTile Algorithm Initialization Initialization Replication Replication Recruitment (Client) (Client) (All Nodes) (All Nodes) (All Nodes) Discovery Discovery Solution (All Nodes) (All Nodes) (Client) Figure 4. Overview of sTile algorithms. W-Sunday, October 28, 12

Privacy Preservation • Each tile being exposed only to a few intermediate • A special tile encodes the solution, but has no knowledge of the input W-Sunday, October 28, 12

Performance Prob. # of Nodes Delay Execution Time Mahjong Private Cluster 20 . 1 sec. A 11 HPCC 19 . 3 sec. PlanetLab 18 . 5 sec. Private Cluster 41 . 6 min. B 11 HPCC 41 . 2 min. PlanetLab 43 . 9 min. Simjong 0ms 65 min. 10ms 57 min. 100ms 64 min. D 1,000,000 500ms 60 min. Gaussian 68 min. Distance-based 59 min. Figure 7. The effect of network delay on system execution time. W-Sunday, October 28, 12

Usage • For Pharma-company: Outsource a protein folding problem to cloud without sharing amino acid sequence private data • Can be use with Mapreduce or Spark W-Sunday, October 28, 12

Comparison • Encryption • [25]Gentry’s result • Secure multi party comm for large data. W-Sunday, October 28, 12

Questions ? 18 W-Sunday, October 28, 12