An analysis of vendor lock-in problem in cloud storage

Big data storage is a considerable challenge for IT organizations. Cloud providers promote using cloud storage for this purpose. However, vendor lock-in problem prevents organizations to migrate toward cloud storage. An evident solution is to distribute redundant data across multiple cloud providers in order to increase the probability of access to data. This paper analyses such a solution that uses erasure coding as a method of distribution. We study the effect of system parameters such as failure rate, redundancy ratio and number of initial block on data access. We show that the probability of access to data varies logarithmically in terms of time and number of initial block. Further increasing redundancy ratio is not always effective.