Toward a Fast and Secure Blockchain Framework for Smart Contracts

继加密货币的成功之后，区块链最近发展成为一个技术平台，为互联网规模的安全、分散和一致的交易分类账提供支持。智能合约部署是许多区块链系统最重要的特性。用户可以开发智能合约来编码任意复杂的交易规则，并将这些合约部署到区块链系统中。然后，合约（即交易规则）由系统的所有参与者忠实地执行和执行，从而消除了未来编码交易的任何潜在对手风险。采用智能合约技术的一个挑战是安全风险。由于智能合约实际上是直接操纵数字资产的程序，因此确保这些合约的正确性非常关键。不幸的是，人类程序员经常会犯错误。智能合约中的编程错误和安全漏洞的后果尤为严重，往往会导致数千万美元的损失。采用的另一个挑战是性能问题。为了确保交易账本的一致性，区块链系统中的共识协议必须以缓慢的块生成方式运行，以避免并发块（即分叉），这被认为对共识安全有害。因此，这些系统对于智能合约的交易吞吐量非常有限。用户通常需要等待很长一段时间才能确认交易。为了应对安全挑战，我们建议设计一套新的工具来保护智能合约，将静态分析与运行时仪器相结合。一个观察结果是，事务执行并不是区块链平台中的瓶颈。瓶颈首先出现在共识协议上，然后是区块链存储层。因此，运行时检测的开销对于其他领域来说通常过于昂贵，实际上对于智能合约来说可以忽略不计。为了解决性能挑战，我们建议设计一种新的共识协议，将块重新组织成直接无环图，以安全地合并并发块。该协议固有地编码了两种策略，一种是用于高性能的乐观策略，另一种是用于确保共识进展以阻止活体攻击的保守策略。我们还建议重新考虑存储层和事务传播的设计，以构建高性能的端到端区块链平台。如果提议的研究取得成功，提议的技术将使智能合约的开发更加安全。这些技术还将允许区块链平台以更高的吞吐量和更快的速度处理智能合约。拟议研究的最终目标是找到全面的解决方案，以释放智能合约技术的全部潜力。

Following the success of the cryptocurrencies, blockchain has recently evolved into a technology platform that powers secure, decentralized, and consistent transaction ledgers at Internet-scale. Smart contract deployment is the most important feature of many blockchain systems. Users can develop smart contracts to encode arbitrarily complicated transaction rules and deploy these contracts to a blockchain system. The contracts (i.e., transaction rules) are then faithfully executed and enforced by all participants of the system, eliminating any potential counter-party risk of the encoded transactions in future. One challenge for the adoption of smart contract techniques is the security risk. Because smart contracts are in fact programs that directly manipulate digital assets, ensuring the correctness of these contracts is very critical. Unfortunately, human programmers often make mistakes and errors. The consequence of programming errors and security vulnerabilities in smart contracts is particularly severe, often leading to tens of millions of dollar losses. Another challenge for the adoption is the performance issue. To ensure the consistency of the transaction ledger, the consensus protocols in blockchain systems have to operate with slow block generation to avoid concurrent blocks (i.e., forks), which are considered harmful for the consensus safety. Therefore these systems have very limited transaction throughputs for smart contracts. Users often have to wait for a long period of time to confirm a transaction. To tackle the security challenge, we propose to design a new set of tools for securing smart contracts, combining the static analysis with runtime instrumentations. One observation is that the transaction execution is not the bottleneck in blockchain platforms. The bottleneck is instead first at the consensus protocol and then at the blockchain storage layer. Therefore the overhead of runtime instrumentation, which is often too expensive for other domains, is in fact negligible for smart contracts. To tackle the performance challenge, we propose to design a novel consensus protocol that reorganizes blocks into direct acyclic graph to safely incorporate concurrent blocks. The protocol inherently encodes two strategies, an optimistic strategy for high performance and a conservative strategy that ensure the consensus progress to thwart liveness attacks. We also propose to revisit the design of the storage layer and the transaction propagation to build a high performance end-to-end blockchain platform. If the proposed research becomes successful, the proposed techniques would make the development of smart contracts significantly more secure. The techniques would also allow blockchain platforms to process smart contracts at a higher throughput and a faster speed. The ultimate goal of the proposed research is to find comprehensive solutions to unleash the full potential of smart contract technologies.