Privacy-Preserving Distributed Encrypted Data Storage and Retrieval Sibi Antony Master Thesis Starting Talk WS12/13 1
Overview Motivation Thesis Goals Methodology Initial Analysis Time Table 2
Motivation Increasing adoption of cloud services Privacy in the data center [ eg : Uniscon ] Storage User db Proxy/Auth Computation Cloud Service 3
Motivation Confidentiality, integrity of data Efficiency and scalability of a cloud system Storage User db Proxy/Auth Computation Cloud Service 4
Thesis Goals 1 Privacy preserving authentication services and user databases Cryptographically securing ACL hierarchy Fully shadowed user data 2 Privacy preserving storage for a distributed system Distributed key generation, encryption and storage 5
By the end of Thesis.. Attempts to answer these questions Key management : Where to store the keys for encryption? Where to perform the encryption operations? Storage Are the user/group, multi-tier information freely accessible? Proof of Storage : Can the user verify the integrity of data? User db Proxy/Auth Computation Are the servers/racks safe from access from an unintended hand? How would it impact the performance and scalability of the system? Cloud Service How to apply my findings on another distributed system? 6
Methodology Study of the Distributed system Understand the security requirements Related work, Compare state-of-the-art Develop prototypes for best-fitting solutions Evaluation in terms of performance and level of security 7
Study: OpenStack storage architecture API Proxy Consistent Hashing Rings Account Object Container Zone1 Zone2 Zone3 n Replicas Storage node storage 1 storage 1 storage 1 Storage servers storage 2 storage 2 storage 2 Consistency servers storage 3 storage 3 storage 3 drive/disk/server storage 4 storage 4 storage 4 8
Initial Analysis : Encryption approaches Proxy-side single point encryption Encryption with a central DB for cryptographic ACL hierarchy Fully distributed encryption Distributed object server encryption Fully Distributed computation Distributed encryption on file chunks Distributed key generation using secure MPC 9
Initial analysis and tests Centralized and Distributed encryption Cost of encryption high in terms of time/performance 10
Time Table A break-up of time alloted for the Thesis work Nov Dec Jan Feb Mar Apr Initial analysis, Design 15th 31st Implementation 1st 15th Tests, Evaluation, Doc 15th Report 15th 11
Thank you! Questions, Discussion. 12
For Your General Interest Www.privacyblog.de Www.idgard.de 13