AT&T Labs Research — Leading Invention, Driving Innovation

Swift is a platform for large-scale visualization of network and transaction-oriented services. It provides interactive views using maps, statistical charts, tables, and network diagrams. The user interface was engineered to run on computers ranging from Powerwall displays to iPhones and other smartphones for remote data access.  Views in Swift are linked both to each other, and to external documents and operations support systems.  For example, an operator can study the impact of a fault within Swift, and easily refer to vendor-supplied documents that explain how to interpret and clear errors.

Its capabilities include:

• ability to show customer impacts of events
• visual querying and drill-down
• runs on standard web clients
• views are generated on the fly from data and metadata
• integrates realtime information with logs
• data manager scales to over 400,000,000 events in a single view
• AT&T vcodex compression can retain years of raw data, e.g. for life cycle analysis

Swift has been applied to a wide range of production services, such as Internet hosting, Internet content distribution, VoIP, WiFi access management, managed security, enterprise call detail analysis. Prototypes were also made using non-communication data: retail point-of-sale transactions, and public health data collected from external news sources.

Swift runs on a proprietary time-series database that can upload and store hundreds of millions of events per day. Typically these are statistics that describe the performance of network and computer system elements. Subject to data retention policies, it is technically feasible to retain detail records of these statistics for a year or longer, using column-oriented compression methods. 

Recent work has been aimed at the visualization of large sets of time series (through a collaboration with Tamara Munzner and Peter McLachlan at the University of British Columbia); inference of faults from network dependency graphs; and a speech-enabled interface. Visitors can see these pilot projects in our lab.

The overall goal of Swift is to integrate and visualize massive data sets, in an innovative, practical software system that provides  transparent data access to AT&T operations and its customers. 

The AT&T Visualizer is the production service offered to enterprise customers that provides visual monitoring and interactive querying of their managed services (Internet hosting, WiFi, managed routers) in Internet Data Centers.  It is provided within the AT&T Business Direct Portal. 

We present LiveRAC, a visualization system that supports the analysis of large collections of system management time-series data consisting of hundreds of parameters across thousands of network devices. LiveRAC provides high information density using a reorderable matrix of charts, with semantic zooming adapting each chart's visual representation to the available space. LiveRAC allows side-by-side visual comparison of arbitrary groupings of devices and parameters at multiple levels of detail. A staged design and development process culminated in the deployment of LiveRAC in a production environment. We conducted an informal longitudinal evaluation of LiveRAC to better understand which proposed visualization techniques were most useful in the target environment.

Visual exploration of massive data sets arising from telecommunication networks and services is a challenge. This paper describes SWIFT-3D, an integrated data visualization and exploration system created at AT&T Labs for large scale network analysis. SWIFT-3D integrates a collection of interactive tools that includes pixel-oriented 2D maps, interactive 3D maps, statistical displays, network topology diagrams and an interactive drill-down query interface. Example applications are described, demonstrating a successful application to analyze unexpected network events (high volumes of unanswered calls), and comparison of usage of an Internet service with voice network traffic and local access coverage.