Blockchain-Based Outsourced Storage Schema in Untrusted Environment
Kun Hao,Junchang Xin,Zhiqiong Wang,Keyan Cao,Guoren Wang
DOI: https://doi.org/10.1109/access.2019.2938578
IF: 3.9
2019-01-01
IEEE Access
Abstract:Outsourced data, as significant service offered by the cloud service provider (CSP), can effectively facilitate the data owner (DO) overcoming the storage limitations on massive data. To ensure the availability of data, DO usually outsources the data replications to multiple CSPs (multi-CSPs) and utilizes a third party metadata management (TPMM) to dominate the metadata of the corresponding replications. However, during the outsourced procedures, DO can hardly confirm the confidence of the TPMM who may take some malicious behaviors to affect the reliability of data. Thus, DO inevitably faces data security issues caused by the over-reliance on the semi-trusted TPMM to manage the metadata of replications. In this paper, we focus on the problem of reliable outsourced data service among multi-CSPs in untrusted environment, that is, how to reliably store and verify the metadata of the data replications in untrusted multi-CSPs environment. To address the problem, we use the novel blockchain technology as a medium to build a trusted outsourced service platform. Moreover, we fully consider the innovative characteristics of blockchain including decentralized architecture, redundancy storage, collective maintenance, and tamper resistant to ensure the data cannot be changed maliciously. We first design a blockchain-based outsourced service framework for storing data replications in untrusted environment, which contains three key layers, that is, storage layer, verification layer, and blockchain layer. Then, we devise a novel concept of verification peer (VP) for maintaining metadata stored by a form of blockchain, and each of which holds the entire blockchain locally to prevent metadata from being maliciously tampered with. Finally, based on the proposed model, we introduce a collaborative algorithm invoked by VPs to store and verify the metadata of replications. We present a completed analysis and conduct extensive experiments on multi-CSPs scenario. The evaluation results demonstrate that our proposed approach achieves superior performance.