Verifiable Distributed Storage Using Blockchain

Abstract

The purpose of the diploma thesis is to develop a data storage system, which provides high performance while ensuring data security. To achieve high performance but also for scalability in relation to the volume of data, we usually resort to distributed approaches such as HDFS or Google GFS. However, as there is no longer a central authority that we trust but in a distributed system nodes can run from different physical entities, the issue of data trust and security arises. To deal with this problem, in this work we use blockchain technology. The ability to execute code in a verifiable way (smart contracts) has led to the adoption of blockchains in a variety of applications beyond cryptocurrencies. The system we developed integrates the Ethereum blockchain into HDFS and ensures that no node can modify / delete files without doing so through the HDFS protocol. The experimental evaluation of the system showed that security does not come for free but we have to sacrifice part of the performance. More specifically, we pay on time an additional x% when importing files and a y% on system availability.