学位论文详细信息
GNuggies: A proposal for hosting resilient stateless services using untrusted nodes
Distributed systems;distributed computing;grid computing;public cloud;incentive mechanism;crowdsourced cloud;serverless computing;untrusted machines;web services;censorship;volunteer compute
Agarwal, Harshit ; Gupta ; Indranil
关键词: Distributed systems;    distributed computing;    grid computing;    public cloud;    incentive mechanism;    crowdsourced cloud;    serverless computing;    untrusted machines;    web services;    censorship;    volunteer compute;   
Others  :  https://www.ideals.illinois.edu/bitstream/handle/2142/101062/AGARWAL-THESIS-2018.pdf?sequence=1&isAllowed=y
美国|英语
来源: The Illinois Digital Environment for Access to Learning and Scholarship
PDF
【 摘 要 】

This thesis outlines a proposal for a serverless cloud compute system hosted on untrusted nodes. We call this proposed system “GNuggies”. It is designed to feel instantly familiar to existing serverless offerings such as AWS Lambda or Azure Cloud Functions. The key difference between GNuggies and existing offerings is that GNuggies proposes leveraging spare compute resources by allowing anyone to contribute nodes into the system. These contributed nodes must be treated as untrusted and this is where the bulk this thesis’s contributions arise:1. A proposed architecture that adapts well understood Distributed Systems concepts to situations involving untrusted nodes and to run in the absence of central authorities.2. A proposed system wherein actors choose to contribute spare compute and are effectively incentivized to do so.3. An incentive structure that makes actors less willing to behave in a malicious manner. This thesis discusses the methods to be used and evaluates their strengths and weaknesses. It also argues that decentralized serverless is a direction that the internet will potentially benefit from moving towards.

【 预 览 】
附件列表
Files Size Format View
GNuggies: A proposal for hosting resilient stateless services using untrusted nodes 662KB PDF download
  文献评价指标  
  下载次数:23次 浏览次数:14次