EAGER: Designing Infrastructure to Probe Distributed System Configurations

网络系统依赖于许多参数来正确配置系统。已知这些参数的不正确设置是造成绝大多数故障、服务质量差和黑客利用网络攻击的原因。同时，诊断错误配置是一个缓慢的、主要是手动的过程，由于对配置参数和系统响应之间的关系以及参数之间的相互依赖性的理解不足，该过程通常需要数天或数周。 该项目将探索自动编写测试、运行测试、收集数据并分析测试所需的功能，以简化查找问题根源的工作。该项目将在数据中心的一些常用服务中构建基本的可诊断性功能，包括域名服务、路由和活动目录沿着适当的访问控制。 该项目还将探索如何在高层次上指定诊断目标，并将其转换为通过输入输出关系连接的基本测试图，并进一步受到访问权限和探测位置的限制。 诊断基础设施将构建和运行测试，收集数据并提供分析数据的系统方法，以尽可能隔离有问题的硬件/软件组件，从而大大加快测试和根本原因分析。该基础设施还将提供存储、排名和重用设计的测试的功能，以使它们随着时间的推移变得更加有效。在该项目下建立的基础设施预计将通过自动化诊断的许多方面大大减少诊断系统的成本，时间和工作量。如果成功的话，这种方法可以应用于新兴的网络物理系统，在这些系统中，错误配置问题可能在本质上更加关键。该项目将增强现有开源软件的基本可诊断性，并开发用于构建和运行复杂测试、引入错误以及收集/分析结果的附加软件。软件和数据将存储在坦普尔的本地档案系统中，并将在奖励期后至少保存三年。 这些数据将通过位于www.example.com的项目网页链接http://www.kkant.net/diagnosisThis奖项反映了NSF的法定使命，并被认为值得通过使用基金会的知识价值和更广泛的影响审查标准进行评估来支持。

Cyber systems depend on a number of parameters to configure the system properly. Incorrect setting of these parameters is known to be responsible for an overwhelming percentage of failures, poor service, and exploitation by hackers for cyberattacks. At the same time, diagnosing misconfigurations is a slow, largely manual process that routinely takes days or weeks because of poor understanding of the relationship between configuration parameters and system's response and interdependencies between the parameters. This project explores the capabilities required to automatically compose tests, run them, collect data, and analyze it to simplify the job of finding the root cause of the problem.The project will build basic diagnosability capabilities in some commonly used services in the data center including domain name service, routing, and active directory along with suitable access control. The project will also explore how the diagnosis goals can be specified at a high-level and translated into a graph of basic tests connected via input-output relationships and further limited by access permissions, and probing locations. The diagnosis infrastructure will build and run the test, collect data and provide a systematic way of analyzing the data to isolate the problematic hardware/software components as much as possible so as to substantially accelerate the testing and root cause analysis. The infrastructure will also provide capabilities to store, rank, and reuse designed tests to make them more effective over time. The infrastructure built under this project is expected to substantially reduce the cost, time, and effort for diagnosing the systems via automation of many aspects of the diagnosis. If successful, the approach can be applied to emerging cyber-physical systems where misconfiguration problems are likely to be even more critical in nature. The project will enhance existing open-source software with basic diagnosability and develop additional software for building and running complex tests, introducing errors, and collecting/analyzing the results. The software and the data will be stored in a local archival system at Temple and will be preserved for at least three years beyond the award period. The data will be linked through the project webpage located at http://www.kkant.net/diagnosisThis award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.