Resource-Efficient Fault and Intrusion Tolerance

基于互联网的服务在我们的社会中发挥着越来越重要的作用，并取代了传统的基础设施。与此同时，这类服务的复杂性以及用于提供这些服务的硬件和软件的复杂性不断增加。尽管采用了先进的开发技术，但这导致了越来越多的系统故障。此外，漏洞允许恶意攻击者从远程攻击这些服务。在最好的情况下，服务只会崩溃，但在其他情况下，服务的完整性可能会受到损害，错误的数据可能会被分发。该项目的目标是为资源高效的拜占庭容错服务提供操作和开发支持。根据项目第一阶段取得的成果，拟议的第二阶段包括两个主要研究领域：(1)如果BFT系统被认为能够容忍灾难，它们必须分布在世界各地。然而，只有当系统能够适应这种环境的特殊条件时，只有当系统具有地理自适应性时，现有资源才能得到有效利用，才能达到所需的性能。在该项目过程中努力采取的措施从动态分配复制品的作用和适应性状态转移到对整个系统进行广泛的重新配置。(2)为了促进这种灵活性，但首先为了有效利用当地资源，第二个研究领域致力于开发能够适应执行这些复制品的平台的特点的复制品。从协议单一关注点的严格分解和新的执行模型出发，设计了一种能够挖掘多核系统和即将到来的非易失性主存的潜力来提高BFT副本的效率的体系结构。总体而言，该项目的第二阶段旨在实现一个资源高效的、地理自适应的、平台感知的BFT系统，该系统能够最大限度地减少甚至克服拜占庭容错需求与现有方案有限的实用性之间的差距。

Internet-based services play a more and more central role in our society and supersede conventional infrastructures. At the same time, the complexity of such services as well as the complexity of the hard- and software employed to provide them continuously increases. Despite advanced development techniques, this leads to a growing number of system failures. Moreover, vulnerabilities allow malicious adversaries to attack these services from remote. In the best case, services simply crash but in other cases, the integrity of services can get compromised and incorrect data can gets distributed.Objective of this project is the provisioning of operating and development support for resource-efficient Byzantine fault-tolerant services. Based on results yielded by the first phase of the project, the proposed second phase comprises two main research areas: (1) If BFT systems are supposed to tolerate catastrophes, they have to be distributed worldwide. However, only if the system is able to adapt to the special conditions of such a setting, only if the system is geo-adaptive, existing resources can be used efficiently and required performance can be achieved. The measures strived for in the course of that project reach from a dynamic assignment of roles for replicas and an adaptive state transfer through to extensive reconfigurations of the whole system. (2) In order to facilitate this kind of flexibility, but first of all in order to make efficient use of local resources, the second research area is dedicated to the development of replicas that are able to adapt to the peculiarities of the platforms executing them. Starting from a strict decomposition of single concerns of the employed protocols and from novel execution models, an architecture is to be devised that can exploit the potential of multi-core systems and upcoming non-volatile main memories to improve the efficiency of BFT replicas.In total, the second phase of this projects aims at a resource-efficient geo-adaptive, platform-aware BFT system that is able to minimize or even overcome the gap between the demand for Byzantine fault tolerance and the confined practicability of current proposals.