Layer-2 Ethernet

Many small to mid-sized service providers have opted to deploy a layer-2 Ethernet network for the provision and transport of Ethernet network services. When implementing a layer-2 (Ethernet) network there are several issues to consider.

Pros

• Low cost devices – with features limited to layer-2 functionality, pure Ethernet equipment is generally more cost effective than equipment that contains higher-level protocol support (IP, etc.)
• Network is a single “domain” – with a single traffic forwarding entity, the configurations and operation complexity incurred with a segmented network is avoided
• Customer interface can be layer-2 or layer-3

Cons

• Spanning tree – slow re-convergence, problematic robustness, and blocked links reduce overall bandwidth capacity
• MAC addresses propagated – with a layer-2 network, MAC addresses are propagated across the network in order to build forwarding tables. In carrier networks, the number of MAC addresses can easily overwhelm typical Ethernet switches.
• No traffic engineering – Ethernet provides limited ability to manage traffic flows.
• Limited QoS and customer traffic segregation features – 802.1P/Q provides basic segregation and quality of service parameters, but cannot provide guaranteed (hard) performance.

Many providers implement a layer-2 network initially - in order to gain experience with the Ethernet services market, business model (cost/revenue calculations) and operations, a low cost infrastructure is a reasonable option. Ultimately though, a pure Ethernet, layer-2 only network is suitable for small networks, but falls short for mid-sized and large network operators. These larger operators require more control and performance from the network infrastructure than Ethernet alone can provide.