BC-Store: A Scalable Design for Blockchain Storage

Abstract

The blockchain technology has obtained significant success in the past decades. However, a serious underlying problem still exists in the blockchain system - data bloating. In the blockchain system, each (full) node must store the full data set in blockchain history, incurring significant storage pressure in the initial synchronization process and the following maintenance of the blockchain system. Data bloating is a challenging problem to be confronted in the immediate future of blockchain. To address this problem, in this paper, we introduce the BC-Store framework that deploys a data accessing model on an IPFS-cluster system to classify the hot and cold blockchain data. The hot data are stored in the local cache, whereas the cold data are stored in the IPFS cluster, thereby substantially shortening the blockchain initial synchronization time and saving a considerable amount of data storage. Empirical experimentation shows that our framework can reduce the local storage size from over 265GB to 4GB with a hit ratio of 77% for Bitcoin without significant performance degradation with the whole data shared in an IPFS cluster.