Mission Aware Data Analysis

该研究计划的目标是开发用于分析异构信息技术（IT）环境中的数据流的技术和工具，以保证任务。挑战在于检测/识别有趣的现象，以便组织的任务/操作即使在故障或攻击条件下也能继续进行。也就是说，IT本身只代表实体的集合，这些实体在部署时可能支持任务保证目标，也可能不支持。只有通过适当的工程实践，才能部署支持系统和技术，以确保IT在支持任务保证方面共同发挥作用。在这种情况下，IT基础设施被认为是任务感知的。考虑到当前IT基础设施的情况，IT计算暗示着云计算和客户端计算（不在云上的计算）共存的未来，并随着数据密集型应用程序的增加而携手发展。随着规模和复杂性的增加，管理员和管理人员分析系统状态变得越来越困难，因为这些系统的性质是不断发展的。因此，如果我们想要任务感知系统，支持当今IT系统的管理系统和技术需要动态地、可预测地监控、分析和适应系统工作负载或环境的快速变化，同时保持对其目标的忠诚。因此，本研究计划的目的是在组织的IT系统从一个层移动到另一个层时自动监视和分析数据流。在这样做的过程中，我们的目标是发现有趣的现象，这些现象可能有助于我们预测故障、性能变化或仅仅是未知/新行为来管理异构系统。为了研究这些问题，我们将采用数据驱动的方法，其中将使用机器学习技术，因为它们有可能将解决方案推广到更广泛的背景下。从长远来看，这一新愿景将为商业、政府和军事需求提供自主和任务感知系统。

The objective of this research program is to develop techniques and tools for analyzing the flow of data in a heterogeneous information technology (IT) environment for mission assurance. The challenge is to detect / identify interesting phenomenon so that the mission / operation of an organization continues even under fault or attack conditions. That is to say, IT in itself only represents a collection of entities, which when deployed, may or may not act to support mission assurance goals. Only through suitable engineering practices can supporting systems and techniques be deployed to ensure that IT collectively functions in support of mission assurance. In such a case the IT infrastructure is said to be mission aware. Given the current landscape of IT infrastructures, IT computing is hinting at a future in which cloud computing and client computing (computing that is not on the cloud) coexist and evolve hand in hand with the increase of data-intensive applications. As the scale and complexity increases, it becomes more and more difficult for administrators and managers to analyze the state of a system because of the continuously evolving nature of these systems. Hence, if we want mission aware systems, the management systems and techniques underpinning today's IT systems need to monitor, analyze and adapt dynamically and predictably to rapid changes in system workload, or environment while remaining true to their objectives. Thus, the purpose of this research program is to automatically monitor and analyze the flow of data on an organization's IT system as it moves from one layer to another. In doing so, we aim to discover interesting phenomenon that might assist us to predict faults, performance changes or simply unknown/new behaviour to manage heterogeneous systems. In order to study these issues, we will employ a data-driven approach where machine-learning techniques will be used, given that they have the potential to generalize solutions to a wider context. In the long term, this new vision provides autonomous and mission aware systems for commercial, governmental and military needs.