Diagnosing network and network application faults remains one of the most challenging and frustrating aspects of using the Internet, particularly for non-technical users, but also for system administrators. A single VoIP application may depend on half a dozen protocols, and basic network access requires numerous network elements, from wireless access points to authorization servers, to function. On a larger scale, researchers have only anecdotal evidence of the most common network problems and their prevalence. This makes it difficult to use actual Internet experiences to improve existing protocols and implementations.
The project proposes a new architecture and system, "Do You See What I See" (DYSWIS), to automate fault finding in residential, mobile, enterprise and data center networks in a systematic, extensible and scalable manner: Using protocol state analysis and packet timing analysis, DYSWIS automatically recognizes when network applications are likely malfunctioning. Leveraging social network graphs, the system finds suitable and mutually-trusting friends whose DYSWIS application can help with finding problems. As necessary, building on a programmable network platform, NetServ, in-network nodes are instantiated and discovered. DYSWIS then attempts to determine the cause of the fault, leveraging network views of other nodes to improve its diagnosis, by remotely invoking active probes at the network, transport and application layer.
DYSWIS collects and aggregates statistics about faults and their resolution, allowing researchers and operators to see trends and determine common sources of problems. DYSWIS is designed to be extensible, so that new standardized and proprietary protocols and applications can be added selectively. The project explores how to scale DYSWIS to the Internet, evaluates its effectiveness, and extends it to use a programmable network platform.
The project enhances undergraduate and graduate teaching, involves undergraduates in research projects, reaches out to industry and standardization organizations, provides tools and large data sets to researchers, and offers a data gathering platform and data for the benefit of Federal and State telecom regulators.
