SCUBA: Focus and Context for Real-Time Mesh Network Health Diagnosis

Abstract

Large-scale wireless metro-mesh networks consisting of hundreds of routers and thousands of clients suffer from a plethora of performance problems. The sheer scale of such networks, the abundance of performance metrics, and the absence of effective tools can quickly overwhelm a network operators’ ability to diagnose these problems. As a solution, we present SCUBA, an interactive focus and context visualization framework for metro-mesh health diagnosis. SCUBA places performance metrics into multiple tiers or contexts, and displays only the topmost context by default to reduce screen clutter and to provide a broad contextual overview of network performance. A network operator can interactively focus on problem regions and zoom to progressively reveal more detailed contexts only in the focal region. We describe SCUBA’s contexts and its planar and hyperbolic views of a nearly 500 node mesh to demonstrate how it eases and expedites health diagnosis. Further, we implement SCUBA on a 15-node testbed, demonstrate its ability to diagnose a problem within a sample scenario, and discuss its deployment challenges in a larger mesh. Our work leads to several future research directions on focus and context visualization and efficient metrics collection for fast and efficient mesh network health diagnosis.