BlockchainDB

With the advent of smart contracts, blockchain platforms, such as Ethereum or Hyperledger Fabric, are being used for many other use cases beyond crypto-currencies. This includes applications in domains such as governmental, healthcare and IoT scenarios. An important aspect that many scenarios have in common, is that blockchains are being used to provide shared data access for parties that do not trust each other. For example, one use case is that a blockchain is used for tracking goods in a supply chain where independent parties log the location of individual goods.

However, while there is a lot of excitement around blockchains in industry, they are still not being used as a shared database (DB) in many real-world scenarios. This has different reasons: First and foremost, a major obstacle is their limited scalability and performance. Recent benchmarks have shown that state-of-the-art blockchain systems such as Ethereum or Hyperledger can only achieve 100sor maximally 1000s of transactions per second, which is often way below the requirements of modern applications. Second, blockchains lack easy-to-use abstractions known from data management systems such as a simple query interface as well as other guarantees like well-defined consistency levels. Instead, blockchains often come with proprietary programming interfaces and require applications to know about the internals of a blockchain to decide on the visibility of updates.

To address the before-mentioned issues, we propose BlockchainDB. BlockchainDB leverages blockchains as the native storage layer and implements an additional database layer on top to enable access to data stored in shared tables. That way, existing blockchain systems can be used (without modification) as a tamper-proof and de-centralized storage. On top of the storage layer, BlockchainDB implements a database layer that provides well known database abstractions and uses additional database techniques to improve the performance of blockchains in the context of data sharing.