Metro Ethernet's third option

SANTA BARBARA, CALIF. - Yet another alternative has emerged for providing SONET-like restoration to metropolitan Ethernet services, which is key to making the LAN technology "carrier-class."

Occam Networks, a developer of broadband loop carrier equipment, last week unveiled its Ethernet Protection Switching (EPS) technology as a standards-based technique for recovering metropolitan Ethernet services in 50 msecs or less. EPS joins the emerging IEEE 802.17 Resilient Packet Ring (RPR) standard and the Metro Ethernet Forum's endorsement of Multi-protocol Label Switching (MPLS) as a third option for equipping Ethernet with SONET-like resiliency.

However, unlike the other two methods, EPS uses the standard IEEE 802.3 Ethernet Media Access Control (MAC) protocol supported by all network equipment. Both RPR and the MPLS protection schemes seek to define new MAC layers.

EPS is available as a standard feature of Occam's broadband loop carrier equipment. But it can also be used by other vendors, Occam says.

EPS, which works over copper and fiber, can be implemented in a variety of network topologies, including ring, star, tree and string.

A string topology is a network in which a node connects with another node, which connects to the following node and so on, in a line. It is assumed that each node performs switching functions between upstream and downstream traffic and on traffic within the node itself.

Ken Twist, program director of broadband access for Ryan Hakim Kent, an industry analysis firm in San Francisco, says Occam's ability to make a network look and behave like SONET could make EPS a big success with large carriers.

"These carriers will not deploy gear unless it can handle voice traffic," Twist says. "The fact that Occam has spent months in the lab trying to get people used to this is a plus. Ethernet comes with a stigma, and to try and change that mindset, to change that paradigm, is going to take some time."

"These guys are the first to propose packetizing the last mile," Twist says. "While that shows great innovation, it could also hurt the adoption process."

How it works

EPS runs in all nodes on a network. It segments traffic into preconfigured flows called path groups using the standard IEEE 802.1Q virtual LAN (VLAN) protocol. Path groups travel in opposite directions on a preconfigured primary physical path. Each path group can contain multiple VLANs, and the network is set up so that no VLAN completes a closed ring.

Each path group is assigned a predefined alternative physical path in the event of a failure. Failures are detected using a "heartbeat" that monitors link operation. Failure by a node to detect two consecutive heartbeats sets off a trigger within the node, and traffic is automatically switched to the alternate physical link.

Occam says heartbeats are sent in 20-msec intervals in typical network designs. Two 20-msec heartbeats missed sequentially, added to a switchover time of less than 10 msecs, results in the 50-msec failover time.

The alternatives

RPR uses two protection mechanisms: packet wrapping and steering.

Packet wrapping means that packets travel in the opposite direction in the event of a failure. This is similar to SONET's protection mechanism.

Steering involves notifying all nodes of a failure and subsequently directing traffic away from a fault. Steering is the default protection mechanism in the most recent draft of the RPR standard, while packet wrapping is optional.

MPLS also has two distinct methods for restoring Ethernet in 50 msecs or less, according to the Metro Ethernet Forum.

The first, called Aggregated Link and Node Protection, addresses local network protection and makes use of MPLS Label Switched Paths. The second, called End-to-End Path Protection, accommodates older network equipment that may not support MPLS.

Back to top

Contact Staff Writer Terri Gimpelson

Other recent articles by Gimpelson