Ethernet has long since broken through the 10 Gigabit barrier, making it sufficiently fast for deployment in MANs and WANs. But network operators and service providers need more than just bandwidth. To become a carrier-grade protocol, Ethernet also has to provide OAM capabilities.

Originally designed as a network protocol for LANs, Ethernet today is a rising technology and is slated to be an essential element in Next Generation Networks (NGNs). But before Ethernet can finally make the grade as a technology for network operators and service providers, one last gap needs to be closed. Equipment providers and network operators are working in standards bodies to equip Ethernet with a capability that is essential in telecommunications. Ethernet can only become a carrier-grade protocol if it incorporates Operations, Administration and Maintenance (OAM) features of the kind supported by TDM networks. In contrast to LANs, service provider networks are large, complex and need to be capable of sustaining large numbers of customers and, not infrequently, several network operators too. When offering customers end-to-end services, today’s methods of LAN management – based for the most part on protocols like SNMP, ping or traceroute – are not sufficient. What’s needed are comprehensive error, network and performance management tools at every network layer. This is the only way that services can be monitored end-to-end and service level agreements (SLAs) verified.

With Ethernet OAM, the exact meaning of the term “Ethernet” is crucial: Ethernet as a service or Ethernet as a transport technology. For Ethernet service providers to operate and monitor their networks and to deliver services with a defined level of quality, they need to be able to manage them at the service layer and at the link layer. This is precisely what the most recent and, in some cases, ongoing standardization efforts have sought to resolve. The IEEE 802.3 working group’s 802.3ah task force has already published an Ethernet OAM protocol for the link layer. However, this only serves to manage Ethernet access lines in “Ethernet in the first mile” (EFM) installations, and is often referred to simply as EFM-OAM.

For carrier end customers like service providers, however, OAM functionalities for the service layer are more important. Here, the main focus is on end-to-end connectivity and service guarantees, and a distinction is made between E-line services (point-to-point and point-to-multipoint) and the E-LAN services (multipoint-to-multipoint) that are important in VLAN applications. The ITU’s Y.1731 working group has already accomplished its task; the IEEE 802.1ag group’s work on connectivity fault management is still in progress. The MEF is also busy working on specifications. Fortunately, all three organizations are working together on Ethernet Service OAM, so it is expected that all of them will approve the final output.

Known as service OAM protocols for simplicity’s sake, these protocols can also be employed when end users are connected on different access technologies. Business users benefit from the ability to provide company locations with seamless connectivity, regardless of the service data rate and the access technology in place. From a network management point of view, it’s unimportant whether the Ethernet services are carried on native Ethernet lines or, say, TDM networks (Ethernet over SDH/PDH/SONET).

End-to-end OAM is also possible when Ethernet services are carried on wireless point-to-point connections and on DSL lines, as well as on metropolitan and wide area networks. With the latter two, the transport technology is again unimportant, because Ethernet switching is supported in the same way as CWDM/DWDM, SDH or IP/MPLS.

An OAM protocol’s most important purpose is to detect network errors and trigger alarms. In the event of an error, it can trigger an SNMP trap or, optionally, switch to an alternative uplink. Statistics and the active alarm log are also updated. Unlike TDM, in which deviations from the constant bit rate are reported as a loss of signal (LOS), Ethernet OAM uses Continuity Check (CC) messages. These are sent over the network as CC OAM frames. The CC OAM frames are sent and received by Ethernet network termination units (E-NTUs) installed on customer premises. If a remote E-NTU does not receive CC OAM frames for a period of 3.5 seconds, a loss of continuity (LOC) is detected and an alarm is triggered to notify personnel that a problem has occurred.

An LOC alarm is canceled if CC OAM frames are received again within a period of 3.5 seconds. In addition, the lost data is replaced with a standard bit pattern. This prevents the alarm from propagating across all subsequent switches and ensures that the operator only receives the alarm from the relevant switch. An AIS OAM message (an Alarm Inhibition Signal or Alarm Indication Signal) indicates that an error has occurred at a prior point along the path. Forward Defect Indication (FDI) indicates to following network elements that a defect has been detected at some previous network element. On bidirectional connections, Reverse Defect Indication (RDI) reports a defect in the opposite direction – for example that a destination could not receive traffic. RDI is generally employed when only portions of the network are managed directly. Unmanaged remote sites cannot directly raise an alarm, so unidirectional loss of continuity to a managed site indicates the defect.

Besides the aforementioned OAM messages that are used during normal operation, there is another class of messages for the initial setup of network equipment and for when equipment restarts after defects have been detected. The most important of these are loopback (LB) messages (LB OAM Request, LB OAM Reply). LB defects are indicated if no LB OAM frames are received within a period of two seconds. There are two general categories of loopback messages: in-service (intrusive) and out-of-service (non-intrusive) messages. In-service LBs are OAM frames that can be sent without disrupting normal operation. These include ping messages. Equipment that receives one of these OAM frames sends it back to its source and does not forward it. All other frames are forwarded as usual. Out-of-service LB commands, by contrast, instruct the interfaces to switch to a special loopback mode that disrupts normal operation.

Network performance parameters typically measured include frame loss, frame delay and frame delay variation. To do this, the mechanisms previously described are extended. CC messages, for example, can be used to determine packet loss by adding a sequence number. By further adding timestamps, CC messages can be used to measure one-way delay. Non-intrusive LB message are used to measure two-way delay. Also, by keeping track of delay measurements, delay variation can be calculated. CC and LB messages similarly can be used for throughput determination in network segments.

For the use of OAM and verification of SLAs based on OAM it is crucial that the Ethernet network termination units (E-NTUs) installed on customer premises clearly distinguish and demarcate boundaries between the customer and service provider networks. This enables high-quality services like VLAN, VPLS, E-line (point-to-point leased line) and E-LAN (multipoint-to-multipoint) to be operated and managed efficiently on heterogeneous infrastructures. With these services, applications like internet access, intranets and extranets, VoIP and video conferencing can be made available to business customers and triple play services can be delivered to private customers over IP DSLAMs.

Manufacturers like RAD Data Communications are already offering solutions today that will, in all likelihood, meet the future Ethernet service OAM standard or will be interoperable with it. These solutions allow monitoring data to be gathered on end-to-end services that enable end customers to verify compliance with service level agreements and providers to demonstrate their fulfillment of SLAs. Thus, OAM is an important prerequisite in the advancement of Ethernet services from easy and inexpensive point-to-point connections into sophisticated multipoint Layer 2 VPNs capable of supporting IP media and communication services in carrier grade quality and creating new possibilities for applications.