SBIR Phase I: Proactive Network Configuration Analysis

Period of Performance: 07/01/2016 - 06/30/2017

$225K

Phase 1 SBIR

Recipient Firm

Intentionet, Inc.
16625 Redmond Way Ste M241 Array
Redmond, WA 98052
Firm POC, Principal Investigator

Abstract

The broader impact/commercial potential of this Small Business Innovation Research (SBIR) Phase I project stems from technology that automatically analyzes network configurations for errors. Computer networks are so tightly woven into the fabric of modern business and society that the delivery of almost all products and services relies on them. Yet networks are notoriously difficult to manage correctly today, and configuration errors that compromise availability, security, and performance are common. Today operators are left to simply wait for bad things to happen and then diagnose and repair the errors as quickly as possible to mitigate the damage. The technology developed in this project will enable organizations to identify network security and availability errors before they are introduced into the running network, thereby saving significant time and money, preventing unauthorized access to customer information, and minimizing down time. The project will also lead to a better understanding of the most prevalent kinds of network configuration errors and how to design networks to prevent them. This Small Business Innovation Research (SBIR) Phase I project will perform the research and development necessary to demonstrate the technical feasibility of a proactive approach to detecting network configuration errors, as instantiated in a software tool. The key innovation underlying the approach is the ability to comprehensively and precisely model and validate the behavior of a network solely by analyzing the network's configuration files. The proposed work has three primary technical goals. First, configuration languages are extremely diverse and complex, so the tool currently only supports features that have been used by the networks to which it has been applied. A key challenge is to augment the logical model underlying the tool to support other features that are used by real-world networks. Second, the tool is currently computationally expensive for large networks. Scalability will be improved by leveraging the structure inherent in network configurations and topologies to perform configuration analysis modularly. Third, to be usable by network operators, the tool must integrate with existing source-control repositories and must provide an expressive interface enabling operators to explore the analysis results. In summary, these research directions will turn the software tool from a research prototype into an expressive, scalable, and usable tool for analyzing real networks.