SBIR Phase II: Proactive Network Configuration Analysis

Intentionet, Inc., Redmond WA

Investigators

Abstract

The broader impact/commercial potential of this Small Business Innovation Research (SBIR) Phase II project stems from technology for automatic network configuration analysis. As ever more devices connect to the Internet and rich services move to the "cloud," both the complexity of computer networks and their reliability requirements are rapidly escalating. It is no wonder that network outages and security breaches are common. Yet another side effect of this complexity, which does not make the headlines but is equally damaging, is that network engineers are understandably fearful of making configuration changes, so networks are unable to evolve at a speed that keeps up with changing business needs. The technology developed in this project will enable network engineers to validate correctness, security, and performance properties of their networks proactively, before errors reach the running network. This technology has the potential to improve the robustness of critical network infrastructure that is widely relied upon, to prevent unauthorized access to resources, and to increase the pace of innovation. The project will also provide insights into the largest "pain points" in modern networks and develop design and analysis techniques to address them. This Small Business Innovation Research (SBIR) Phase II project further develops the Batfish network configuration analysis technology. Interactions with pilot customers as well as many interviews with potential customers identified the needs of the marketplace and how the Batfish technology can best be adapted to meet those needs. As a result, the specific goal of this project is to seamlessly aid network engineers in validating network behavior during the policy design phase. Drawing inspiration from how software is developed today, the company will extend Batfish to support continuous integration of network configurations and develop a series of analyses that can find errors in network configurations with minimal input from the network engineers. The anticipated outcome of these research thrusts is a technology that allows network engineers to easily understand and gain confidence in their proposed network designs and to iterate these designs more quickly. The project will be driven by continued interactions with several pilot customers.