Effectiveness of Oblivious RAM in cloud storage services

MASTER Assignment

Effectiveness of Oblivious RAM in cloud storage services

Type : Master M-CS

Period: Jan, 2022 - Aug, 2022

Student : Brandt, J.H. (Jacco, Student M-CS)

Date Final project: Aug 29, 2022

Thesis

Supervisors:

C.H.M. van den Bogaard, MSc (Deloitte)

Abstract:

Cloud storage services provide their users with external storage space that can be used to save files on an external server. Such storage services usually support encryption techniques because of privacy and security considerations. Regular encryption techniques can guarantee that the content of these files remains confidential, however these methods cannot hide the access patterns of the files. Oblivious RAM(ORAM) algorithms are techniques that are used to protect both the content and access pattern of stored data. In this paper we measure how to transform these files into blocks suitable for usage by ORAM algorithms efficiently, and compare three of the most well-known ORAM algorithms to verify which are best suited for usage in cloud storage services. Our results show that both Square-root ORAM and PathORAM are suitable for usage in cloud storage services, but their usage scenarios differ. Square-root ORAM is best suited for performing automated backups, because it has a high worst-case performance which regularly causes requests to take a long time. PathORAM is best suited in situations where cloud storage services are used to reduce the local storage by storing files externally, because it has a constant cost performance ensuring stable and fast reponse to requests.