Using Ethernet for differentiated services

One of the characteristics of Carrier Ethernet that make it a truly carrier-class technology is its support for differentiated services. Service providers can take advantage of this to offer a wide range of Ethernet services that can be delivered end to end, whether they operate in the retail, wholesale or even mobile space.

Unsurprisingly, however, different operators in different sectors have different network requirements when it comes to deploying differentiated Ethernet services, which also means different solutions and best practices, to include areas of first-layer aggregation, demarcation and Ethernet access. 

As such, there are several key differences operators should be aware of when it comes to running Ethernet services for wholesale, retail and mobile backhaul networks.
Key differences 

Retail customers will typically be located both in-region and out-of-region. This means relying on wholesalers for out of region access and could conceivably require negotiating service contracts with a number of wholesalers in order to service a retail customer. This can be time consuming. Carrier Ethernet exchanges, which could ameliorate some of the inter-connect difficulties and speed the process, are beginning to gain traction but are not universally available. The last mile in any event is probably over different physical infrastructures, but the end user quality of experience must remain the same. 

The retail operator must also ensure SLAs to the end user and verify it vis a vis its customers. There are a couple of scenarios for this. The Carrier Ethernet (CE) demarcation device or NID could be deployed back-to-back by both the wholesaler and retail operator, or the wholesale operator would manage the service end-to-end in its own region but allow the retail operator access to the PM KPIs or the wholesale operator would offer a service VPN to the retail operator from the single CE demarcation device.

Retail customers also require multiple services ?four to five, on average ?with a mixture of CIR and EIR bandwidth. If some of the retail branches have high capacity needs but can only be reached over bonded copper or DSL bonding then sophisticated traffic management schemes are required in the CE demark device to avoid congestion and service impairment.

Also network requirements vary according to retail customers?applications. Are they connecting data centers, departmental LANs? How much bandwidth is required per site? How scalable is the network? Perhaps the retail operator will provide a mix of Layer 2 and Layer 3 services. Is the traffic bursty? Symmetric or asymmetric? Financial service customers require ultra low latency, for example. All of these will determine the retail operators?service offering to their customer.

Wholesale services require higher capacity in the access '1G/10G/40G' and typically sell their retail customers guaranteed bandwidth between their own PoP and the retail service providers nearest the PoP. The wholesale operator doesn't care what is running in the pipe. They must also ensure high availability and resiliency, including EVC protection (which is also relevant for certain retail applications like financial services). The wholesale operator's SLA will look one way to their retail service provider customer. 

Mobile operators (or more precisely wholesale mobile backhaul providers) will need to support various or multiple timing over packet technologies since the base stations or network might not be supporting the same synchronization capabilities (sync-Ethernet or 1588v2). The mobile operator must also be able to connect the base stations in a self-healing ring topology, which means support for G.8032 Ethernet Ring Protection Switching or close the loop in a drop-and-insert or hub and spoke topology.    
Standalone vs multi-service carrier 

It's also worth pointing out the different requirements for deploying differentiated Ethernet services for a multi-service carrier (i.e. one that offers retail, wholesale and mobile), as opposed to a standalone operator offering only one such service.

The multi-service carrier has different operating centers, and in some ways each operating center can be treated as a separate segment. However, at the board level the operator will want to leverage synergies to cut costs and benefit from economies of scale. 

For this reason, infrastructure could be shared, but from a service perspective, the services will remain separate. This means the retail arm of a multi-service carrier will probably deploy it own demarcation device, while the wholesale arm will do the same. Depending on regulatory requirements and commercial objectives, the wholesale arm of an incumbent will also offer services to other service providers. 

The mobile operator may require a cell-site gateway with pseudowire support, while the wholesale arm will use a mobile demarcation device to separate its network from the mobile network. 

The standalone operator can specialize in a particular infrastructure or service niche.

Best practices for QoS, COS and SLAs

All of the vendors of CE demarcation devices support (at least on paper) the standard set of connectivity verification, service turn up, performance monitoring, and fault management tools defined by the ITU, IEEE and IETF for end-to-end SLA assurance. 

Not all of them, however, support these processes in hardware. Using a hardware-powered CE demarc device provides for rapid detection of service deterioration to ensure under 50 msec switchover, highly accurate frame loss measurement capabilities (down to a single frame loss) as well as frame delay and jitter, multiple monitoring of simultaneous OAM flows as well as wire speed data transfer. 

The sum of these capabilities is a demarc that guarantees precise, real-time SLA assurance capabilities over multiple sites. Another best practice is to provide traffic shaping per EVC CoS. This is particularly important for supporting premium service differentiation in retail offerings and avoiding congestion in bandwidth limited access networks. 
Planning for future traffic growth

One key issue operators face is how to effectively plan for future traffic growth and scalability, particularly with video expected to be the major traffic driver in the next several years. 

In terms of best practices, service providers will need to deploy a non-blocking 1G (or even 10G for corporate headquarters) wire speed demarc at the customer premises with sophisticated two-way traffic management capabilities to the user and the network, giving priority to video applications and avoiding congestion, as well as assuring high availability and resilience (G.8031 EVC protection) through the network. The first level aggregator should support multicast and IGMP snooping as well as G.8032 and G.8031. 

Ilan Seidner is director of marketing communicatinos for RAD Data Communications