With the advancement of cloud computing and internet of things, data are usually stored on distributed computers and these data may risk being lost or stolen. In this article, we consider a common case where the entirety of the data is partitioned into several parts and each data part can be allocated to one or more computers. In the case where a computer fails, all the data parts on it are lost. Before the failure of any computer, the data parts may also be stolen by hackers. The basic model of computer failure and computer intrusion resulting in the theft of all the data parts on the computer is considered first. Then, the case is extended to a general model where computer failure, as well as data part corruption and theft caused by hacking are embedded. It is essential to study the reliability of distributed storage systems considering both data loss and data theft, which can be a basis for decision making on system structure optimization. In this article, a multi-valued decision diagram–based approach is developed to quantitatively evaluate system reliability for both models considering the time-dependence property of sequential events. The proposed method is applicable to systems where the random time to failure, theft, or corruption follows arbitrary distributions including the commonly used exponential distributions. Illustrative examples are provided to validate the proposed method.
Amenability, extreme amenability, model-theoretic stability, and dependence property in integral logic
