Method for Estimating Locations of Service Problem Causes in Service Function Chaining

Abstract

We propose a method for estimating the locations of the causes of failure or deterioration in quality in the service function chains (SFCs) of datacenter networks by using tests packets. Our method is especially effective in service-quality deterioration where no alarms are notified from physical and virtual components. This method models SFC configurations by creating a correlation ID, which identifies a correlation between physical network elements and virtualized elements of the SFC components to enable us to find the components common to all SFCs that are either in failure or deteriorating. We consider such common components as locations where service problems can occur in SFCs. As a result of applying this method to SFCs in datacenter networks, it drastically reduced the number of candidate causes and computation time.