Sentinel: in Case of the Untrustworthy Behaviors Inside the Clouds

Dong Cui,Chuanyi Liu,Meiqi Yang,Jincui Yang
DOI: https://doi.org/10.1007/978-3-662-47401-3_13
2015-01-01
Abstract:Since compute cloud is the most important part of IAAS applications, there are a number of security issues associated with it, including the threat from untrustworthy administrators who may compromise the users' system without authorization and escape responsibility by deleting logs. Based on previous publication, ways to prevent destruction from administrators mainly fall into three broad categories: SSO(Single Sign On), administration rights distribution, log analyzing. However, any of these methods have strengths and weaknesses. In this document, we come up with a new method "Sentinel", which combines double-check and the log mechanism and can ensure the security of the system without weakening administrators' privileges. In our Sentinel, the administrators can only enter the management domain through a unified entrance. The Sentinel is able to detect and intersect destructive instruction and at the same time keep operation logs which are transparent to the administrators.
What problem does this paper attempt to address?