Compiling Abstract Specifications into Concrete Systems - Bringing Order to the Cloud
by Ian Unruh, Alexandru G. Bardas, Rui Zhuang, Xinming Ou, and Scott A. DeLoach.
Abstract: Currently, there are no suitable abstractions that allow cloud users to define the structure and dependency of the services in a cloud-based IT system. As a result, cloud users either have to manage the low-level details of the cloud services directly, such as in IaaS, or resort to SaaS or PaaS where much or part of the services are managed by the cloud provider but are less flexible to customize to better suit user needs. We propose a high-level abstraction called the requirement model for defining cloud-based IT systems. It captures the important aspects of a system's structure, such as service dependencies, without considering the low-level details such as operating systems or application configurations. The requirement model separates the cloud customer's concern of what the system does, from the system engineer's concern of how to implement it. We further develop a "compilation" process that automatically translates a requirement model into a concrete system based on pre-defined and reusable knowledge units. This higher-level specification and the associated compilation process allows repeatable deployment of cloud-based IT systems, supports more reliable system management, and enables implementing the same requirement in multiple ways. We demonstrate the practicality of this approach in the ANCOR (Automated eNterprise network COmpileR) framework, which takes a requirement model and generates an IT system based on that specification. Our current implementation targets OpenStack and uses Puppet to configure the cloud instances, although it could work with other cloud platforms and configuration management solutions as well.
Multi-Factor Authentication for More Resilient Distributed
Storage in Wireless Networks
Scott Bell, Eugene Vasserman, and Daniel Andresen.
Abstract: Modern military units derive great tactical advantage from secure real-time sharing of data such as maps, images and orders among soldiers. However, distributing this information in real time in a combat situation creates significant risk: when using a mobile communication network, the adversary may capture one or more mobile devices, gaining access to this data and endangering the entire unit. While these devices are generally tamper-resistant and require a login, this would not deter a well-funded and motivated attacker. In this work, we present a protocol which significantly reduces the adversary's window of opportunity for attack by incorporating distributed content storage and revocable authentication for users and individual devices without increasing the difficulty of soldier-device interaction. To further limit an adversary's ability to access this data, file requests must contain fresh authentication information from both a trusted user and device. We analyze the benefits and trade-offs of this protocol both theoretically and through tests on real-world mobile devices, and find that the computation, response, and battery overhead are acceptable purely in software, and can be greatly reduced with inexpensive hardware acceleration. Simulations indicate the probability of data loss is significantly reduced over systems which only require a user secret (password or PIN).