Bringing Decentralized Search to Decentralized Services

Bringing Decentralized Search to Decentralized Services
复制标题

DOI:
--
复制
发表时间:
2021
期刊:
--
影响因子:
--
通讯作者:
Mingyu Li;Jinhao Zhu;Tianxu Zhang;Cheng Tan;Yubin Xia;Sebastian Angel;Haibo Chen
Mingyu Li;Jinhao Zhu;Tianxu Zhang;Cheng Tan;Yubin Xia;Sebastian Angel;Haibo Chen
中科院分区:
其他
文献类型:
--
作者:
Mingyu Li;Jinhao Zhu;Tianxu Zhang;Cheng Tan;Yubin Xia;Sebastian Angel;Haibo Chen

文献摘要

相似文献

最近出现的去中心化服务,如Steemit和OpenBazaar,提供了负责任的社交媒体和在线电子商务。然而,尽管它们本身是去中心化的,但它们仍然依赖于集中式或不安全的分布式搜索引擎来进行最终用户导航,这使得这些服务容易受到审查和隐私侵犯。这项工作提出了D E S Escort,第一个分散的搜索引擎,保证分散服务的搜索结果的完整性和隐私。D E S Eccessor使用可信的硬件来构建一个流水线网络,这是一小部分外包工作的飞地原语。由于每个lambda只有一个局部视图,D E S Escort引入了一种新的见证机制,该机制暗示了lambda的传递性,以使这种分散的模型可以外部验证。D E S Etrance支持查询的快速验证,并可以容忍分散式故障。我们实现了一个去中心化的社交媒体和在线市场的D E S Escort原型。评估表明,D E S Escort可以在去中心化环境中水平扩展,每天在234 GB的真实数据集上支持3200万个请求,其中包含8200万个文档。
Recent decentralized services such as Steemit and OpenBazaar have emerged to provide accountable social media and online e-commerce. However, despite being decentralized themselves, they still rely on a centralized or an insecure distributed search engine for end-user navigation, which makes such services vulnerable to censorship and privacy violation. This work presents D E S EARCH , the first decentralized search engine that guarantees the integrity and privacy of search results for decentralized services. D E S EARCH uses trusted hardware to build a network of pipelined lambdas , an enclave primitive for a small piece of outsourced job. Since each lambda only has a local view, D E S EARCH introduces a novel witness mechanism that implies the transitivity of lambdas to make this decentralized model externally verifiable. D E S EARCH enables fast verification for queries and can tolerate decentralized failures. We implement a D E S EARCH prototype for a decentralized social media and online market. Evaluation shows that D E S EARCH can scale horizontally in a decentralized environment and support 32 million requests per day on a 234 GB real-world dataset with 82M documents.