TC: Small: Improving System Security through Virtual Layered File Systems

台式计算机运行许多不同的应用程序，由于应用程序之间缺乏隔离，任何一个应用程序的损坏都可能危及整个桌面。恢复受损的桌面仍然是一项耗时的任务，通常需要清除所有内容并重新安装系统。这些安全问题带来了根本性的挑战，因为从金融交易到医疗记录，一切都依赖台式电脑。为了解决这些问题，我们正在创建新的虚拟分层文件系统（VLFS）技术来提高系统安全性。与传统的单一文件系统不同，VLFS动态地将一组软件层组合到桌面的单个文件系统视图中。对一个层的更改与对另一个层的更改是隔离和解耦的。VLFS动态组合特性实现了强大且易于使用的安全功能。我们正在使用vlfse来构建一个架构，以便在管理大量异构配置的机器时有效地部署安全补丁，并加速系统从安全漏洞中恢复。我们还使用vlfse来开发透明的桌面应用程序故障遏制体系结构，该体系结构可以有效地限制漏洞利用造成的损害，从而实现快速恢复，同时与传统桌面系统一样易于使用。本提案的结果将为未来的计算机创新提供基础，为用户系统提供改进的系统安全性。因为我们使用的是工业标准操作系统、二进制应用程序和补丁发行版，所以我们的结果将直接适用于商业世界。

Desktop computers run many different applications, thecompromise of any one of which can compromise the entire desktop giventhe lack of isolation among applications. Recovering a compromiseddesktop remains a time consuming task, which typically requires wipingeverything and reinstalling the system from scratch. These security issues pose fundamental challenges as desktop computers arerelied on for everything from financial transactions tomedical records. To address these problems, we are creating novelvirtual layered file system (VLFS) technologies to improve system security. Unlike a traditional file system which is a monolithic entity, a VLFS dynamically composes together a set of software layersinto a single file system view for a desktop. Changes to one layerare isolated and decoupled from changes to another. The VLFS dynamiccomposition feature enables powerful and easy-to-use securityfunctionality. We are using VLFSes to build an architecture to enablesecurity patches to be deployed effectively when managing largenumbers of heterogeneously configured machines, and to speed systemrecovery from security exploits. We are also using VLFSes to developa transparent desktop application fault containment architecture thatis effective at limiting the damage from exploits to enable quickrecovery while being as easy to use as a traditional desktop system.The results of this proposal will provide a foundation for futurecomputer innovations to provide improved system security for users'systems. Because we are working with industry-standard operatingsystems and binary application and patch distributions, our resultswill be directly applicable to the commercial world.