Network Segmentation Fault Detection and Remediation

A network fault that occurs such that the two ends of a shared network cannot communicate with each other, but each side still has external network connectivity, is referred to as a network segmentation fault.

An everRun system provides a network segmentation detection mechanism that places the active VM on the node that has the most external network connectivity when the system detects this fault. As part of this feature, the everRun system constantly sends UDP packets over the business network interface between the active and stand-by nodes. The system's network segmentation logic detects a fault when this packet flow is interrupted while both sides still have an active network link. In this fault scenario, both nodes still have active network connections, so the fault lies in a switch that is external to the everRun system.

When this case is detected, the everRun system handles the fault based on logic that determines which side has better external connectivity. The everRun system makes this fault-handling decision by continually monitoring incoming broadcast/multicast traffic to determine which node has the most incoming traffic. In this fault case, if the VM is not already active on the node with the most incoming network traffic, the everRun system fails the VM network over to this node. The fault detection feature requires no user configuration since it is basing the decision on traffic that normally occurs on any system.

Related Topics

Network Architecture Overview