(Accedian Networks') EtherNEWS Blog

Carrier Ethernet Services offer a carrot to customers with EIR (or Excess Information Rate) – possible bonus bandwidth, network congestion permitting. EIR is for packets that live in the “yellow light district”- a best-effort zone where throughput may or may not be available, and where there are no guarantees they’ll ever get to the other end.

Serious business and wireless backhaul applications require dedicated bandwidth to function predictably, provisioned as the Committed Information Rate (CIR) of an Ethernet Virtual Circuit (EVC). The MEF labels this as “Green” traffic, packets that have a confirmed ticket to ride the network from end-to-end. For services with EIR enabled, traffic exceeding the CIR but less than the EIR get a yellow tag – meaning these packets enter the network, but may get dropped at any time, at any hop along the way if congestion is encountered. Traffic exceeding the EIR is simply dropped – these are truly packets in the red-zone.

Despite the chancy nature of EIR, customers see value in it, especially for less important services. Carrier Ethernet switches can assign traffic priorities through hierarchical rate limiting (bandwidth policing) to ensure that delay-sensitive or critical traffic gets all green lights, while lower priority traffic potentially face yellow or red. EIR is cheap compared to CIR, and has great benefit when everyday tasks like data center access, email and Internet browsing run up bandwidth in bursts during peak usage hours. Service providers offering Carrier Ethernet E-Lines and ELANs sweeten their appeal with EIR on the menu; legacy data services including ATM and T1s are unable to accommodate.

The EIR Gifted?

Many service providers introducing unmanaged Ethernet services find themselves uncertain about network performance. Lacking the visibility offered by Ethernet Operations, Administration & Maintenance (OAM), and the bandwidth policing capabilities of MEF-certified Carrier Ethernet network elements, these are the EIR Gifted, meaning they have plenty of EIR. In fact their whole network is EIR since they can’t provide any guarantees! Services come in one flavor, EIR only. You take your chances with critical applications and hope the network’s not too oversubscribed when you need it most. For those with a sense of adventure, the EIR Gifted are the perfect suitor. Cavaet: you may find you get what you pay for.

The EIR Challenged

At the other end of the spectrum are operators offering Ethernet over TDM infrastructures. The very nature of SONET / SDH is that a dedicated circuit with a fixed capacity is established for each point-to-point connection. This is the ultimate CIR connection – with low latency and fully committed bandwidth assured by the underlying nature of the transport technology. For demanding applications, it’s an excellent service delivery architecture. But for most enterprises, it might be just a bit too good. A mix of CIR and EIR is normally sufficient and more cost effective, but when it comes to EIR, these providers hesitate to offer it. Why? With bandwidth provisioned in dedicated increments, they must commit equally valuable network capacity to deliver either CIR or EIR. From an engineering perspective these services are the same. These are the EIR Challenged. They would like to offer EIR but end up delivering CIR anyway (at a discount).

Billing further complicates the matter. How do you charge for EIR usage if the pipe is always available? Most customers expect to pay only for the EIR they use. What used to be a simple matter of billing dedicated bandwidth now requires usage metering, non-TDM-like back-office and billing systems. And then there’s the chance that the customer figures out that the EIR also offers the same performance as the CIR band and is always available. Upselling capacity, one of the many benefits of Ethernet, just got harder.

Virtual EIR

Short of ripping out their network and deploying native Ethernet or MPLS, what are the EIR Challenged to do when their customers come knocking? Many service providers are now creating “Virtual EIR” using traffic shaping. By conditioning a customer’s traffic as it enters the network, providers can accommodate bursts, improve bandwidth efficiency and greatly reduce packet loss, all while maintaining a simple, CIR-based service. The result is similar to EIR, but no additional network capacity is needed. The customer sees increased availability, and can burst into the network without penalty for short periods of time.

Unlike real EIR, which has a hard limit for bandwidth peaks, shaping takes a softer approach. Bursts well beyond a “real EIR” can be buffered and retransmitted when capacity frees up, and only when shaping buffers are full will packets be dropped. Complementing the Carrier Ethernet traffic coloring scheme, priority, delay-sensitive traffic is not only given the green light, it can bypass shaping queues altogether. This means that real-time applications like VoIP and video see no added latency or jitter from the shaping mechanism, while ensuring they have 100% assured transit within the available CIR.

This virtualized EIR offers several advantages to both the customer and the service provider, while maintaining a simple billing and service provisioning model. Traffic shaping can be offered as a service in its own right, and monetized in a number of ways. The most common is a flat monthly fee added to the customer’s invoice, sometimes with a setup charge that covers the cost of the NID or switch that will perform the shaping. As NIDs also offer performance monitoring, some operators sell these devices directly to their customers, comparing them to a set-top box for HDTV: they enable premium, SLA-grade, fully assured and shaped services – a step up from their standard offering, and a commitment to the customer that they’ll get what they pay for. EIR Challenged no longer. From now on it’s EIR On Tap.