GRuB: Gas-Efficient Blockchain Storage via Workload-Adaptive Data Replication
Modern Blockchains support the execution of user programs, called smart contracts. As a trusted computing platform, smart contracts bring decentralization, computation integrity, open access and information transparency to average users on the Internet. However, running smart-contract programs leads to high costs, known as Gas. Such costs prevent the use of smart contracts in data-intensive application scenarios, such as high-frequency trading and transparency logging. This paper addresses the Gas-based cost effectiveness in the most consuming layer of a smart contract, namely data storage. We present GRuB, a dynamic data-replication framework that monitors the smart-contract workload and makes online replication decisions. A new online algorithm is proposed that provides constant-bounded 'competitiveness' in Gas. To further save Gas, the workload monitor and decision maker are run off the Blockchain and with security against the forging of workload trace being monitored. A GRuB prototype is built, including a smart-contract component on Ethereum and an off-chain middleware on top of Google LevelDB. The cost evaluation under the YCSB workloads shows that GRuB can converge quickly to changing workloads and save Gas significantly compared with static replication schemes. Two case studies are conducted for data-intensive applications, including high-frequency trading and transparency logging, in which running GRuB leads to affordable Gas.
READ FULL TEXT