Book excerpt: At the current state, traditional Ethernet is not suitable for the new era of Ethernet applications, specifically Metro Ethernet Network and Industrial Ethernet Network. With the new requirements for Quality of Service and the wide spread adoption of Virtual Private LAN Service, the landscape of the link layer is rapidly changing. However, over the years, Ethernet has proven that a decentralized system is robust to failures. Therefore, our proposed solution is a decentralized system to support a robust Carrier Ethernet with focuses on: 1. Characterized the key weakness in the topology management of Ethernet. 2. Proposed an scalable efficient end-to-end topology management in MEN. 3. Supported multi-path forwarding in Ethernet via distributed overlay nodes. 4. Provisioned QoS in Ethernet through a system of Multiple Spanning Tree. 5. Virtualized network resources to reduce failure recovery to under 1ms. We propose three approaches in managing the decentralized network: 1. Overlay: A failure can be quickly averted by diverting the stream of traffic over to the peer super node. Using the super nodes, we can reduce the convergence time by eliminating the need to have the control messages to propagate throughout the global topology. The QoS requirements can be communicated just between the super nodes to guarantee flows end to end without the expensive procedure of assessing each individual node for each possible path. 2. Spanning Tree: We explored Multiple Spanning Tree to exploit the benefits of sending traffic on multiple Spanning Trees concurrently in case of load imbalance and failures. In addition, the Spanning Trees can be used as a way to assert QoS management such as traffic policing, admission control, and guarantee of service for high priority traffic. We showed how the allocation of Spanning Tree can affect the performance of each class of service. 3. Virtualization: The physical topology is virtualized into multiple logical systems serving as supporting LAN where each one is independent of another. The topologies do not require a global arbitrator to coordinate them. Each node in the network is micromanaged to direct the traffic in response to a network failure within a few microseconds.