Protection at the MAC and physical layers: self-healing rings

A ring network is a network topology where all nodes are attached to the same set of physical links. Each link forms a loop. In counter rotating ring topologies, all links are unidirectional and traffic flows in one direction on one half of the links, and in the reverse direction on the other half. Self-healing rings are particular counter rotating ring networks which perform rerouting as follows. In normal operation, traffic is sent from a source to a destination in one direction only. If a link fails, then the other direction is used to reach the destination such that the failed link is avoided. Self-healing rings require expensive specific hardware and waste up to half of the available bandwidth to provide full redundancy. On the other hand, lower layer protection mechanisms are the fastest rerouting mechanisms available as self-healing rings can reroute traffic in less than 50 ms. In this section, we present four MAC and physical rerouting mechanisms which all rely on a counter rotating ring topology: SONET UPSR and BLSR Automatic Protection Switching, FDDI protection switching, and RPR Intelligent Protection Switching.

SONET is a physical layer technology for optical transmission [11] [12]. In SONET, protection with self-healing rings is called ``Automatic Protection Switching'' (APS [10]) and comes in two flavors. The first one, Unidirectional Path-Switched Ring architecture (UPSR, see Figure 2.2), benefits from 1+1 protection. In 1+1 protection, two rings are used. A source injects exactly the same traffic in reverse directions on both rings. The destination receives the same data on each ring, but takes into account traffic from one ring only. On link failure, the receiver detects the increase of the bit error rate or the absence of traffic on one of the rings, and then decides to take into account the traffic from the other ring. The SONET standards specify that the service interruption time should not exceed 50 ms, which is low enough for the outage to be unnoticeable by customers who participate in a live conversation where voice is carried over a SONET network. While service recovery with this technique meets the 50 ms goal, SONET UPSR requires a substantial amount of dedicated backup resources as half of the links are used for path restoration purpose only.

Figure 2.2: SONET self-healing ring: Unidirectional Path-Switched Ring architecture. UPSR achieves 1+1 protection.

Figure 2.3: SONET self-healing ring: Bidirectional Link-Switched Ring architecture. BLSR achieves 1:1 protection.

The second protecting scheme, Bidirectional Link-Switched Ring architecture (BLSR, see Figure 2.3), benefits from 1:1 protection. In 1:1 protection, every link can carry both regular traffic and backup traffic at the same time and thus does not require dedicated backup links. On a link failure, the node upstream of the failed link wraps traffic from one ring to another ring in the reverse direction so that traffic still can reach its destination. BLSR is as fast as Unidirectional Path-Switched protection and does not waste as many resources, as there is no notion of dedicated primary and backup link [11].

The MAC layer provides the means for IP to send packets over a local area network. Fiber Distributed Data Interface (FDDI [4]) implements at the MAC layer a protection mechanism that is similar to SONET BLSR. FDDI runs over dual counter rotating rings. In normal operation, traffic is sent on one ring only. Like BLSR, FDDI wraps paths when a link failure is detected and uses the second ring only as a backup ring. Therefore, FDDI implements 1+1 protection and requires full link redundancy.

Figure 2.4: Intelligent Protection Switching with a Resilient Packet Ring. IPS is an enhancement of BLSR.

Resilient Packet Ring is a more recent MAC protocol designed to run on multiple counter-rotating rings (see Figure 2.4(a)) [37]. In RPR, path protection is called Intelligent Protection Switching (IPS). IPS can be viewed as an enhanced SONET BLSR mechanism. Indeed, when a link failure occurs, traffic is first wrapped exactly like SONET BLSR does (Figure 2.4(b)). The emitting node is notified of the failure and changes the ring on which it sends traffic (Figure 2.4(c)). The new path taken by packets is therefore shorter than the wrapped path, resulting in both shorter delays for packets and a better utilization of the available resources.

The lower layer rerouting mechanisms are fast because the nodes that detect the failure perform themselves instantaneously the switchover step, bypassing the notification step. The total repair time is therefore reduced to the detection time ( $T_{repair} = T_{detect}$).